package gocv

/*
#include <stdlib.h>
#include "imgproc.h"
*/
import "C"
import (
	"image"
	"image/color"
	"reflect"
	"unsafe"
)

func getPoints(pts *C.Point, l int) []C.Point {
	h := &reflect.SliceHeader{
		Data: uintptr(unsafe.Pointer(pts)),
		Len:  l,
		Cap:  l,
	}
	return *(*[]C.Point)(unsafe.Pointer(h))
}

// ArcLength calculates a contour perimeter or a curve length.
//
// For further details, please see:
//
// https://docs.opencv.org/master/d3/dc0/group__imgproc__shape.html#ga8d26483c636be6b35c3ec6335798a47c
//
func ArcLength(curve []image.Point, isClosed bool) float64 {
	cPoints := toCPoints(curve)
	arcLength := C.ArcLength(cPoints, C.bool(isClosed))
	return float64(arcLength)
}

// ApproxPolyDP approximates a polygonal curve(s) with the specified precision.
//
// For further details, please see:
//
// https://docs.opencv.org/master/d3/dc0/group__imgproc__shape.html#ga0012a5fdaea70b8a9970165d98722b4c
//
func ApproxPolyDP(curve []image.Point, epsilon float64, closed bool) (approxCurve []image.Point) {
	cCurve := toCPoints(curve)

	cApproxCurve := C.ApproxPolyDP(cCurve, C.double(epsilon), C.bool(closed))
	defer C.Points_Close(cApproxCurve)

	cApproxCurvePoints := getPoints(cApproxCurve.points, int(cApproxCurve.length))

	approxCurve = make([]image.Point, cApproxCurve.length)
	for i, cPoint := range cApproxCurvePoints {
		approxCurve[i] = image.Pt(int(cPoint.x), int(cPoint.y))
	}
	return approxCurve
}

// ConvexHull finds the convex hull of a point set.
//
// For further details, please see:
// https://docs.opencv.org/master/d3/dc0/group__imgproc__shape.html#ga014b28e56cb8854c0de4a211cb2be656
//
func ConvexHull(points []image.Point, hull *Mat, clockwise bool, returnPoints bool) {
	cPoints := toCPoints(points)
	C.ConvexHull(cPoints, hull.p, C.bool(clockwise), C.bool(returnPoints))
}

// ConvexityDefects finds the convexity defects of a contour.
//
// For further details, please see:
// https://docs.opencv.org/master/d3/dc0/group__imgproc__shape.html#gada4437098113fd8683c932e0567f47ba
//
func ConvexityDefects(contour []image.Point, hull Mat, result *Mat) {
	cPoints := toCPoints(contour)
	C.ConvexityDefects(cPoints, hull.p, result.p)
}

// CvtColor converts an image from one color space to another.
// It converts the src Mat image to the dst Mat using the
// code param containing the desired ColorConversionCode color space.
//
// For further details, please see:
// http://docs.opencv.org/master/d7/d1b/group__imgproc__misc.html#ga4e0972be5de079fed4e3a10e24ef5ef0
//
func CvtColor(src Mat, dst *Mat, code ColorConversionCode) {
	C.CvtColor(src.p, dst.p, C.int(code))
}

// EqualizeHist normalizes the brightness and increases the contrast of the image.
//
// For further details, please see:
// https://docs.opencv.org/master/d6/dc7/group__imgproc__hist.html#ga7e54091f0c937d49bf84152a16f76d6e
func EqualizeHist(src Mat, dst *Mat) {
	C.EqualizeHist(src.p, dst.p)
}

// CalcHist Calculates a histogram of a set of images
//
// For futher details, please see:
// https://docs.opencv.org/master/d6/dc7/group__imgproc__hist.html#ga6ca1876785483836f72a77ced8ea759a
func CalcHist(src []Mat, channels []int, mask Mat, hist *Mat, size []int, ranges []float64, acc bool) {
	cMatArray := make([]C.Mat, len(src))
	for i, r := range src {
		cMatArray[i] = r.p
	}

	cMats := C.struct_Mats{
		mats:   (*C.Mat)(&cMatArray[0]),
		length: C.int(len(src)),
	}

	chansInts := []C.int{}
	for _, v := range channels {
		chansInts = append(chansInts, C.int(v))
	}
	chansVector := C.struct_IntVector{}
	chansVector.val = (*C.int)(&chansInts[0])
	chansVector.length = (C.int)(len(chansInts))

	sizeInts := []C.int{}
	for _, v := range size {
		sizeInts = append(sizeInts, C.int(v))
	}
	sizeVector := C.struct_IntVector{}
	sizeVector.val = (*C.int)(&sizeInts[0])
	sizeVector.length = (C.int)(len(sizeInts))

	rangeFloats := []C.float{}
	for _, v := range ranges {
		rangeFloats = append(rangeFloats, C.float(v))
	}
	rangeVector := C.struct_FloatVector{}
	rangeVector.val = (*C.float)(&rangeFloats[0])
	rangeVector.length = (C.int)(len(rangeFloats))

	C.CalcHist(cMats, chansVector, mask.p, hist.p, sizeVector, rangeVector, C.bool(acc))
}

// CalcBackProject calculates the back projection of a histogram.
//
// For futher details, please see:
// https://docs.opencv.org/3.4/d6/dc7/group__imgproc__hist.html#ga3a0af640716b456c3d14af8aee12e3ca
func CalcBackProject(src []Mat, channels []int, hist Mat, backProject *Mat, ranges []float64, uniform bool) {
	cMatArray := make([]C.Mat, len(src))
	for i, r := range src {
		cMatArray[i] = r.p
	}

	cMats := C.struct_Mats{
		mats:   (*C.Mat)(&cMatArray[0]),
		length: C.int(len(src)),
	}

	chansInts := []C.int{}
	for _, v := range channels {
		chansInts = append(chansInts, C.int(v))
	}
	chansVector := C.struct_IntVector{}
	chansVector.val = (*C.int)(&chansInts[0])
	chansVector.length = (C.int)(len(chansInts))

	rangeFloats := []C.float{}
	for _, v := range ranges {
		rangeFloats = append(rangeFloats, C.float(v))
	}
	rangeVector := C.struct_FloatVector{}
	rangeVector.val = (*C.float)(&rangeFloats[0])
	rangeVector.length = (C.int)(len(rangeFloats))

	C.CalcBackProject(cMats, chansVector, hist.p, backProject.p, rangeVector, C.bool(uniform))
}

// HistCompMethod is the method for Histogram comparison
// For more information, see https://docs.opencv.org/master/d6/dc7/group__imgproc__hist.html#ga994f53817d621e2e4228fc646342d386
type HistCompMethod int

const (
	// HistCmpCorrel calculates the Correlation
	HistCmpCorrel HistCompMethod = 0

	// HistCmpChiSqr calculates the Chi-Square
	HistCmpChiSqr = 1

	// HistCmpIntersect calculates the Intersection
	HistCmpIntersect = 2

	// HistCmpBhattacharya applies the HistCmpBhattacharya by calculating the Bhattacharya distance.
	HistCmpBhattacharya = 3

	// HistCmpHellinger applies the HistCmpBhattacharya comparison. It is a synonym to HistCmpBhattacharya.
	HistCmpHellinger = HistCmpBhattacharya

	// HistCmpChiSqrAlt applies the Alternative Chi-Square (regularly used for texture comparsion).
	HistCmpChiSqrAlt = 4

	// HistCmpKlDiv applies the Kullback-Liebler divergence comparison.
	HistCmpKlDiv = 5
)

// CompareHist Compares two histograms.
//
// For further details, please see:
// https://docs.opencv.org/master/d6/dc7/group__imgproc__hist.html#gaf4190090efa5c47cb367cf97a9a519bd
func CompareHist(hist1 Mat, hist2 Mat, method HistCompMethod) float32 {
	return float32(C.CompareHist(hist1.p, hist2.p, C.int(method)))
}

// BilateralFilter applies a bilateral filter to an image.
//
// Bilateral filtering is described here:
// http://www.dai.ed.ac.uk/CVonline/LOCAL_COPIES/MANDUCHI1/Bilateral_Filtering.html
//
// BilateralFilter can reduce unwanted noise very well while keeping edges
// fairly sharp. However, it is very slow compared to most filters.
//
// For further details, please see:
// https://docs.opencv.org/master/d4/d86/group__imgproc__filter.html#ga9d7064d478c95d60003cf839430737ed
//
func BilateralFilter(src Mat, dst *Mat, diameter int, sigmaColor float64, sigmaSpace float64) {
	C.BilateralFilter(src.p, dst.p, C.int(diameter), C.double(sigmaColor), C.double(sigmaSpace))
}

// Blur blurs an image Mat using a normalized box filter.
//
// For further details, please see:
// https://docs.opencv.org/master/d4/d86/group__imgproc__filter.html#ga8c45db9afe636703801b0b2e440fce37
//
func Blur(src Mat, dst *Mat, ksize image.Point) {
	pSize := C.struct_Size{
		width:  C.int(ksize.X),
		height: C.int(ksize.Y),
	}

	C.Blur(src.p, dst.p, pSize)
}

// BoxFilter blurs an image using the box filter.
//
// For further details, please see:
// https://docs.opencv.org/master/d4/d86/group__imgproc__filter.html#gad533230ebf2d42509547d514f7d3fbc3
//
func BoxFilter(src Mat, dst *Mat, depth int, ksize image.Point) {
	pSize := C.struct_Size{
		height: C.int(ksize.X),
		width:  C.int(ksize.Y),
	}
	C.BoxFilter(src.p, dst.p, C.int(depth), pSize)
}

// SqBoxFilter calculates the normalized sum of squares of the pixel values overlapping the filter.
//
// For further details, please see:
// https://docs.opencv.org/master/d4/d86/group__imgproc__filter.html#ga045028184a9ef65d7d2579e5c4bff6c0
//
func SqBoxFilter(src Mat, dst *Mat, depth int, ksize image.Point) {
	pSize := C.struct_Size{
		height: C.int(ksize.X),
		width:  C.int(ksize.Y),
	}
	C.SqBoxFilter(src.p, dst.p, C.int(depth), pSize)
}

// Dilate dilates an image by using a specific structuring element.
//
// For further details, please see:
// https://docs.opencv.org/master/d4/d86/group__imgproc__filter.html#ga4ff0f3318642c4f469d0e11f242f3b6c
//
func Dilate(src Mat, dst *Mat, kernel Mat) {
	C.Dilate(src.p, dst.p, kernel.p)
}

// DistanceTransformLabelTypes are the types of the DistanceTransform algorithm flag
type DistanceTransformLabelTypes int

const (
	// DistanceLabelCComp assigns the same label to each connected component of zeros in the source image
	// (as well as all the non-zero pixels closest to the connected component).
	DistanceLabelCComp DistanceTransformLabelTypes = 0

	// DistanceLabelPixel assigns its own label to each zero pixel (and all the non-zero pixels closest to it).
	DistanceLabelPixel
)

// DistanceTransformMasks are the marsk sizes for distance transform
type DistanceTransformMasks int

const (
	// DistanceMask3 is a mask of size 3
	DistanceMask3 DistanceTransformMasks = 0

	// DistanceMask5 is a mask of size 3
	DistanceMask5

	// DistanceMaskPrecise is not currently supported
	DistanceMaskPrecise
)

// DistanceTransform Calculates the distance to the closest zero pixel for each pixel of the source image.
//
// For further details, please see:
// https://docs.opencv.org/master/d7/d1b/group__imgproc__misc.html#ga8a0b7fdfcb7a13dde018988ba3a43042
//
func DistanceTransform(src Mat, dst *Mat, labels *Mat, distType DistanceTypes, maskSize DistanceTransformMasks, labelType DistanceTransformLabelTypes) {
	C.DistanceTransform(src.p, dst.p, labels.p, C.int(distType), C.int(maskSize), C.int(labelType))
}

// Erode erodes an image by using a specific structuring element.
//
// For further details, please see:
// https://docs.opencv.org/master/d4/d86/group__imgproc__filter.html#gaeb1e0c1033e3f6b891a25d0511362aeb
//
func Erode(src Mat, dst *Mat, kernel Mat) {
	C.Erode(src.p, dst.p, kernel.p)
}

// RetrievalMode is the mode of the contour retrieval algorithm.
type RetrievalMode int

const (
	// RetrievalExternal retrieves only the extreme outer contours.
	// It sets `hierarchy[i][2]=hierarchy[i][3]=-1` for all the contours.
	RetrievalExternal RetrievalMode = 0

	// RetrievalList retrieves all of the contours without establishing
	// any hierarchical relationships.
	RetrievalList = 1

	// RetrievalCComp retrieves all of the contours and organizes them into
	// a two-level hierarchy. At the top level, there are external boundaries
	// of the components. At the second level, there are boundaries of the holes.
	// If there is another contour inside a hole of a connected component, it
	// is still put at the top level.
	RetrievalCComp = 2

	// RetrievalTree retrieves all of the contours and reconstructs a full
	// hierarchy of nested contours.
	RetrievalTree = 3

	// RetrievalFloodfill lacks a description in the original header.
	RetrievalFloodfill = 4
)

// ContourApproximationMode is the mode of the contour approximation algorithm.
type ContourApproximationMode int

const (
	// ChainApproxNone stores absolutely all the contour points. That is,
	// any 2 subsequent points (x1,y1) and (x2,y2) of the contour will be
	// either horizontal, vertical or diagonal neighbors, that is,
	// max(abs(x1-x2),abs(y2-y1))==1.
	ChainApproxNone ContourApproximationMode = 1

	// ChainApproxSimple compresses horizontal, vertical, and diagonal segments
	// and leaves only their end points.
	// For example, an up-right rectangular contour is encoded with 4 points.
	ChainApproxSimple = 2

	// ChainApproxTC89L1 applies one of the flavors of the Teh-Chin chain
	// approximation algorithms.
	ChainApproxTC89L1 = 3

	// ChainApproxTC89KCOS applies one of the flavors of the Teh-Chin chain
	// approximation algorithms.
	ChainApproxTC89KCOS = 4
)

// BoundingRect calculates the up-right bounding rectangle of a point set.
//
// For further details, please see:
// https://docs.opencv.org/3.3.0/d3/dc0/group__imgproc__shape.html#gacb413ddce8e48ff3ca61ed7cf626a366
//
func BoundingRect(contour []image.Point) image.Rectangle {
	cContour := toCPoints(contour)
	r := C.BoundingRect(cContour)
	rect := image.Rect(int(r.x), int(r.y), int(r.x+r.width), int(r.y+r.height))
	return rect
}

// BoxPoints finds the four vertices of a rotated rect. Useful to draw the rotated rectangle.
//
// For further Details, please see:
// https://docs.opencv.org/3.3.0/d3/dc0/group__imgproc__shape.html#gaf78d467e024b4d7936cf9397185d2f5c
//
func BoxPoints(rect RotatedRect, pts *Mat) {

	rPoints := toCPoints(rect.Contour)

	rRect := C.struct_Rect{
		x:      C.int(rect.BoundingRect.Min.X),
		y:      C.int(rect.BoundingRect.Min.Y),
		width:  C.int(rect.BoundingRect.Max.X - rect.BoundingRect.Min.X),
		height: C.int(rect.BoundingRect.Max.Y - rect.BoundingRect.Min.Y),
	}

	rCenter := C.struct_Point{
		x: C.int(rect.Center.X),
		y: C.int(rect.Center.Y),
	}

	rSize := C.struct_Size{
		width:  C.int(rect.Width),
		height: C.int(rect.Height),
	}

	r := C.struct_RotatedRect{
		pts:          rPoints,
		boundingRect: rRect,
		center:       rCenter,
		size:         rSize,
		angle:        C.double(rect.Angle),
	}

	C.BoxPoints(r, pts.p)
}

// ContourArea calculates a contour area.
//
// For further details, please see:
// https://docs.opencv.org/3.3.0/d3/dc0/group__imgproc__shape.html#ga2c759ed9f497d4a618048a2f56dc97f1
//
func ContourArea(contour []image.Point) float64 {
	cContour := toCPoints(contour)
	result := C.ContourArea(cContour)
	return float64(result)
}

type RotatedRect struct {
	Contour      []image.Point
	BoundingRect image.Rectangle
	Center       image.Point
	Width        int
	Height       int
	Angle        float64
}

// MinAreaRect finds a rotated rectangle of the minimum area enclosing the input 2D point set.
//
// For further details, please see:
// https://docs.opencv.org/3.3.0/d3/dc0/group__imgproc__shape.html#ga3d476a3417130ae5154aea421ca7ead9
//
func MinAreaRect(points []image.Point) RotatedRect {
	cPoints := toCPoints(points)
	result := C.MinAreaRect(cPoints)

	defer C.Points_Close(result.pts)
	pArray := result.pts.points
	pLength := int(result.pts.length)

	pHdr := reflect.SliceHeader{
		Data: uintptr(unsafe.Pointer(pArray)),
		Len:  pLength,
		Cap:  pLength,
	}
	sPoints := *(*[]C.Point)(unsafe.Pointer(&pHdr))

	points4 := make([]image.Point, pLength)
	for j, pt := range sPoints {
		points4[j] = image.Pt(int(pt.x), int(pt.y))
	}

	return RotatedRect{
		Contour:      points4,
		BoundingRect: image.Rect(int(result.boundingRect.x), int(result.boundingRect.y), int(result.boundingRect.x)+int(result.boundingRect.width), int(result.boundingRect.y)+int(result.boundingRect.height)),
		Center:       image.Pt(int(result.center.x), int(result.center.y)),
		Width:        int(result.size.width),
		Height:       int(result.size.height),
		Angle:        float64(result.angle),
	}
}

// MinEnclosingCircle finds a circle of the minimum area enclosing the input 2D point set.
//
// For further details, please see:
// https://docs.opencv.org/3.4/d3/dc0/group__imgproc__shape.html#ga8ce13c24081bbc7151e9326f412190f1
func MinEnclosingCircle(points []image.Point) (x, y, radius float32) {
	cPoints := toCPoints(points)
	cCenterPoint := C.struct_Point2f{}
	var cRadius C.float
	C.MinEnclosingCircle(cPoints, &cCenterPoint, &cRadius)
	x, y = float32(cCenterPoint.x), float32(cCenterPoint.y)
	radius = float32(cRadius)
	return x, y, radius
}

// FindContours finds contours in a binary image.
//
// For further details, please see:
// https://docs.opencv.org/3.3.0/d3/dc0/group__imgproc__shape.html#ga17ed9f5d79ae97bd4c7cf18403e1689a
//
func FindContours(src Mat, mode RetrievalMode, method ContourApproximationMode) [][]image.Point {
	ret := C.FindContours(src.p, C.int(mode), C.int(method))
	defer C.Contours_Close(ret)

	cArray := ret.contours
	cLength := int(ret.length)
	cHdr := reflect.SliceHeader{
		Data: uintptr(unsafe.Pointer(cArray)),
		Len:  cLength,
		Cap:  cLength,
	}
	sContours := *(*[]C.Points)(unsafe.Pointer(&cHdr))

	contours := make([][]image.Point, cLength)
	for i, pts := range sContours {
		pArray := pts.points
		pLength := int(pts.length)
		pHdr := reflect.SliceHeader{
			Data: uintptr(unsafe.Pointer(pArray)),
			Len:  pLength,
			Cap:  pLength,
		}
		sPoints := *(*[]C.Point)(unsafe.Pointer(&pHdr))

		points := make([]image.Point, pLength)
		for j, pt := range sPoints {
			points[j] = image.Pt(int(pt.x), int(pt.y))
		}
		contours[i] = points
	}

	return contours
}

//ConnectedComponentsAlgorithmType specifies the type for ConnectedComponents
type ConnectedComponentsAlgorithmType int

const (
	// SAUF algorithm for 8-way connectivity, SAUF algorithm for 4-way connectivity.
	CCL_WU ConnectedComponentsAlgorithmType = 0

	// BBDT algorithm for 8-way connectivity, SAUF algorithm for 4-way connectivity.
	CCL_DEFAULT = 1

	// BBDT algorithm for 8-way connectivity, SAUF algorithm for 4-way connectivity
	CCL_GRANA = 2
)

// ConnectedComponents computes the connected components labeled image of boolean image.
//
// For further details, please see:
// https://docs.opencv.org/master/d3/dc0/group__imgproc__shape.html#gaedef8c7340499ca391d459122e51bef5
//
func ConnectedComponents(src Mat, labels *Mat) int {
	return int(C.ConnectedComponents(src.p, labels.p, C.int(8), C.int(MatTypeCV32S), C.int(CCL_DEFAULT)))
}

// ConnectedComponents computes the connected components labeled image of boolean image.
//
// For further details, please see:
// https://docs.opencv.org/master/d3/dc0/group__imgproc__shape.html#gaedef8c7340499ca391d459122e51bef5
//
func ConnectedComponentsWithParams(src Mat, labels *Mat, conn int, ltype MatType,
	ccltype ConnectedComponentsAlgorithmType) int {
	return int(C.ConnectedComponents(src.p, labels.p, C.int(conn), C.int(ltype), C.int(ccltype)))
}

// ConnectedComponentsTypes are the connected components algorithm output formats
type ConnectedComponentsTypes int

const (
	//The leftmost (x) coordinate which is the inclusive start of the bounding box in the horizontal direction.
	CC_STAT_LEFT = 0

	//The topmost (y) coordinate which is the inclusive start of the bounding box in the vertical direction.
	CC_STAT_TOP = 1

	// The horizontal size of the bounding box.
	CC_STAT_WIDTH = 2

	// The vertical size of the bounding box.
	CC_STAT_HEIGHT = 3

	// The total area (in pixels) of the connected component.
	CC_STAT_AREA = 4

	CC_STAT_MAX = 5
)

// ConnectedComponentsWithStats computes the connected components labeled image of boolean
// image and also produces a statistics output for each label.
//
// For further details, please see:
// https://docs.opencv.org/master/d3/dc0/group__imgproc__shape.html#ga107a78bf7cd25dec05fb4dfc5c9e765f
//
func ConnectedComponentsWithStats(src Mat, labels *Mat, stats *Mat, centroids *Mat) int {
	return int(C.ConnectedComponentsWithStats(src.p, labels.p, stats.p, centroids.p,
		C.int(8), C.int(MatTypeCV32S), C.int(CCL_DEFAULT)))
}

// ConnectedComponentsWithStats computes the connected components labeled image of boolean
// image and also produces a statistics output for each label.
//
// For further details, please see:
// https://docs.opencv.org/master/d3/dc0/group__imgproc__shape.html#ga107a78bf7cd25dec05fb4dfc5c9e765f
//
func ConnectedComponentsWithStatsWithParams(src Mat, labels *Mat, stats *Mat, centroids *Mat,
	conn int, ltype MatType, ccltype ConnectedComponentsAlgorithmType) int {
	return int(C.ConnectedComponentsWithStats(src.p, labels.p, stats.p, centroids.p, C.int(conn),
		C.int(ltype), C.int(ccltype)))
}

// TemplateMatchMode is the type of the template matching operation.
type TemplateMatchMode int

const (
	// TmSqdiff maps to TM_SQDIFF
	TmSqdiff TemplateMatchMode = 0
	// TmSqdiffNormed maps to TM_SQDIFF_NORMED
	TmSqdiffNormed = 1
	// TmCcorr maps to TM_CCORR
	TmCcorr = 2
	// TmCcorrNormed maps to TM_CCORR_NORMED
	TmCcorrNormed = 3
	// TmCcoeff maps to TM_CCOEFF
	TmCcoeff = 4
	// TmCcoeffNormed maps to TM_CCOEFF_NORMED
	TmCcoeffNormed = 5
)

// MatchTemplate compares a template against overlapped image regions.
//
// For further details, please see:
// https://docs.opencv.org/master/df/dfb/group__imgproc__object.html#ga586ebfb0a7fb604b35a23d85391329be
//
func MatchTemplate(image Mat, templ Mat, result *Mat, method TemplateMatchMode, mask Mat) {
	C.MatchTemplate(image.p, templ.p, result.p, C.int(method), mask.p)
}

// Moments calculates all of the moments up to the third order of a polygon
// or rasterized shape.
//
// For further details, please see:
// https://docs.opencv.org/master/d3/dc0/group__imgproc__shape.html#ga556a180f43cab22649c23ada36a8a139
//
func Moments(src Mat, binaryImage bool) map[string]float64 {
	r := C.Moments(src.p, C.bool(binaryImage))

	result := make(map[string]float64)
	result["m00"] = float64(r.m00)
	result["m10"] = float64(r.m10)
	result["m01"] = float64(r.m01)
	result["m20"] = float64(r.m20)
	result["m11"] = float64(r.m11)
	result["m02"] = float64(r.m02)
	result["m30"] = float64(r.m30)
	result["m21"] = float64(r.m21)
	result["m12"] = float64(r.m12)
	result["m03"] = float64(r.m03)
	result["mu20"] = float64(r.mu20)
	result["mu11"] = float64(r.mu11)
	result["mu02"] = float64(r.mu02)
	result["mu30"] = float64(r.mu30)
	result["mu21"] = float64(r.mu21)
	result["mu12"] = float64(r.mu12)
	result["mu03"] = float64(r.mu03)
	result["nu20"] = float64(r.nu20)
	result["nu11"] = float64(r.nu11)
	result["nu02"] = float64(r.nu02)
	result["nu30"] = float64(r.nu30)
	result["nu21"] = float64(r.nu21)
	result["nu12"] = float64(r.nu12)
	result["nu03"] = float64(r.nu03)

	return result
}

// PyrDown blurs an image and downsamples it.
//
// For further details, please see:
// https://docs.opencv.org/master/d4/d86/group__imgproc__filter.html#gaf9bba239dfca11654cb7f50f889fc2ff
//
func PyrDown(src Mat, dst *Mat, ksize image.Point, borderType BorderType) {
	pSize := C.struct_Size{
		height: C.int(ksize.X),
		width:  C.int(ksize.Y),
	}
	C.PyrDown(src.p, dst.p, pSize, C.int(borderType))
}

// PyrUp upsamples an image and then blurs it.
//
// For further details, please see:
// https://docs.opencv.org/master/d4/d86/group__imgproc__filter.html#gada75b59bdaaca411ed6fee10085eb784
//
func PyrUp(src Mat, dst *Mat, ksize image.Point, borderType BorderType) {
	pSize := C.struct_Size{
		height: C.int(ksize.X),
		width:  C.int(ksize.Y),
	}
	C.PyrUp(src.p, dst.p, pSize, C.int(borderType))
}

// MorphologyDefaultBorder returns "magic" border value for erosion and dilation.
// It is automatically transformed to Scalar::all(-DBL_MAX) for dilation.
//
// For further details, please see:
// https://docs.opencv.org/master/d4/d86/group__imgproc__filter.html#ga94756fad83d9d24d29c9bf478558c40a
//
func MorphologyDefaultBorderValue() Scalar {
	var scalar C.Scalar = C.MorphologyDefaultBorderValue()
	return NewScalar(float64(scalar.val1), float64(scalar.val2), float64(scalar.val3), float64(scalar.val4))
}

// MorphologyEx performs advanced morphological transformations.
//
// For further details, please see:
// https://docs.opencv.org/master/d4/d86/group__imgproc__filter.html#ga67493776e3ad1a3df63883829375201f
//
func MorphologyEx(src Mat, dst *Mat, op MorphType, kernel Mat) {
	C.MorphologyEx(src.p, dst.p, C.int(op), kernel.p)
}

// MorphologyExWithParams performs advanced morphological transformations.
//
// For further details, please see:
// https://docs.opencv.org/master/d4/d86/group__imgproc__filter.html#ga67493776e3ad1a3df63883829375201f
//
func MorphologyExWithParams(src Mat, dst *Mat, op MorphType, kernel Mat, iterations int, borderType BorderType) {
	pt := C.struct_Point{
		x: C.int(-1),
		y: C.int(-1),
	}
	C.MorphologyExWithParams(src.p, dst.p, C.int(op), kernel.p, pt, C.int(iterations), C.int(borderType))
}

// MorphShape is the shape of the structuring element used for Morphing operations.
type MorphShape int

const (
	// MorphRect is the rectangular morph shape.
	MorphRect MorphShape = 0

	// MorphCross is the cross morph shape.
	MorphCross = 1

	// MorphEllipse is the ellipse morph shape.
	MorphEllipse = 2
)

// GetStructuringElement returns a structuring element of the specified size
// and shape for morphological operations.
//
// For further details, please see:
// https://docs.opencv.org/master/d4/d86/group__imgproc__filter.html#gac342a1bb6eabf6f55c803b09268e36dc
//
func GetStructuringElement(shape MorphShape, ksize image.Point) Mat {
	sz := C.struct_Size{
		width:  C.int(ksize.X),
		height: C.int(ksize.Y),
	}

	return newMat(C.GetStructuringElement(C.int(shape), sz))
}

// MorphType type of morphological operation.
type MorphType int

const (
	// MorphErode operation
	MorphErode MorphType = 0

	// MorphDilate operation
	MorphDilate = 1

	// MorphOpen operation
	MorphOpen = 2

	// MorphClose operation
	MorphClose = 3

	// MorphGradient operation
	MorphGradient = 4

	// MorphTophat operation
	MorphTophat = 5

	// MorphBlackhat operation
	MorphBlackhat = 6

	// MorphHitmiss operation
	MorphHitmiss = 7
)

// BorderType type of border.
type BorderType int

const (
	// BorderConstant border type
	BorderConstant BorderType = 0

	// BorderReplicate border type
	BorderReplicate = 1

	// BorderReflect border type
	BorderReflect = 2

	// BorderWrap border type
	BorderWrap = 3

	// BorderReflect101 border type
	BorderReflect101 = 4

	// BorderTransparent border type
	BorderTransparent = 5

	// BorderDefault border type
	BorderDefault = BorderReflect101

	// BorderIsolated border type
	BorderIsolated = 16
)

// GaussianBlur blurs an image Mat using a Gaussian filter.
// The function convolves the src Mat image into the dst Mat using
// the specified Gaussian kernel params.
//
// For further details, please see:
// http://docs.opencv.org/master/d4/d86/group__imgproc__filter.html#gaabe8c836e97159a9193fb0b11ac52cf1
//
func GaussianBlur(src Mat, dst *Mat, ksize image.Point, sigmaX float64,
	sigmaY float64, borderType BorderType) {
	pSize := C.struct_Size{
		width:  C.int(ksize.X),
		height: C.int(ksize.Y),
	}

	C.GaussianBlur(src.p, dst.p, pSize, C.double(sigmaX), C.double(sigmaY), C.int(borderType))
}

// Sobel calculates the first, second, third, or mixed image derivatives using an extended Sobel operator
//
// For further details, please see:
// https://docs.opencv.org/master/d4/d86/group__imgproc__filter.html#gacea54f142e81b6758cb6f375ce782c8d
//
func Sobel(src Mat, dst *Mat, ddepth, dx, dy, ksize int, scale, delta float64, borderType BorderType) {
	C.Sobel(src.p, dst.p, C.int(ddepth), C.int(dx), C.int(dy), C.int(ksize), C.double(scale), C.double(delta), C.int(borderType))
}

// SpatialGradient calculates the first order image derivative in both x and y using a Sobel operator.
//
// For further details, please see:
// https://docs.opencv.org/master/d4/d86/group__imgproc__filter.html#ga405d03b20c782b65a4daf54d233239a2
//
func SpatialGradient(src Mat, dx, dy *Mat, ksize int, borderType BorderType) {
	C.SpatialGradient(src.p, dx.p, dy.p, C.int(ksize), C.int(borderType))
}

// Laplacian calculates the Laplacian of an image.
//
// For further details, please see:
// https://docs.opencv.org/master/d4/d86/group__imgproc__filter.html#gad78703e4c8fe703d479c1860d76429e6
//
func Laplacian(src Mat, dst *Mat, dDepth int, size int, scale float64,
	delta float64, borderType BorderType) {
	C.Laplacian(src.p, dst.p, C.int(dDepth), C.int(size), C.double(scale), C.double(delta), C.int(borderType))
}

// Scharr calculates the first x- or y- image derivative using Scharr operator.
//
// For further details, please see:
// https://docs.opencv.org/master/d4/d86/group__imgproc__filter.html#gaa13106761eedf14798f37aa2d60404c9
//
func Scharr(src Mat, dst *Mat, dDepth int, dx int, dy int, scale float64,
	delta float64, borderType BorderType) {
	C.Scharr(src.p, dst.p, C.int(dDepth), C.int(dx), C.int(dy), C.double(scale), C.double(delta), C.int(borderType))
}

// MedianBlur blurs an image using the median filter.
//
// For further details, please see:
// https://docs.opencv.org/master/d4/d86/group__imgproc__filter.html#ga564869aa33e58769b4469101aac458f9
//
func MedianBlur(src Mat, dst *Mat, ksize int) {
	C.MedianBlur(src.p, dst.p, C.int(ksize))
}

// Canny finds edges in an image using the Canny algorithm.
// The function finds edges in the input image image and marks
// them in the output map edges using the Canny algorithm.
// The smallest value between threshold1 and threshold2 is used
// for edge linking. The largest value is used to
// find initial segments of strong edges.
// See http://en.wikipedia.org/wiki/Canny_edge_detector
//
// For further details, please see:
// http://docs.opencv.org/master/dd/d1a/group__imgproc__feature.html#ga04723e007ed888ddf11d9ba04e2232de
//
func Canny(src Mat, edges *Mat, t1 float32, t2 float32) {
	C.Canny(src.p, edges.p, C.double(t1), C.double(t2))
}

// CornerSubPix Refines the corner locations. The function iterates to find
// the sub-pixel accurate location of corners or radial saddle points.
//
// For further details, please see:
// https://docs.opencv.org/master/dd/d1a/group__imgproc__feature.html#ga354e0d7c86d0d9da75de9b9701a9a87e
//
func CornerSubPix(img Mat, corners *Mat, winSize image.Point, zeroZone image.Point, criteria TermCriteria) {
	winSz := C.struct_Size{
		width:  C.int(winSize.X),
		height: C.int(winSize.Y),
	}

	zeroSz := C.struct_Size{
		width:  C.int(zeroZone.X),
		height: C.int(zeroZone.Y),
	}

	C.CornerSubPix(img.p, corners.p, winSz, zeroSz, criteria.p)
	return
}

// GoodFeaturesToTrack determines strong corners on an image. The function
// finds the most prominent corners in the image or in the specified image region.
//
// For further details, please see:
// https://docs.opencv.org/master/dd/d1a/group__imgproc__feature.html#ga1d6bb77486c8f92d79c8793ad995d541
//
func GoodFeaturesToTrack(img Mat, corners *Mat, maxCorners int, quality float64, minDist float64) {
	C.GoodFeaturesToTrack(img.p, corners.p, C.int(maxCorners), C.double(quality), C.double(minDist))
}

// GrabCutMode is the flag for GrabCut algorithm.
type GrabCutMode int

const (
	// GCInitWithRect makes the function initialize the state and the mask using the provided rectangle.
	// After that it runs the itercount iterations of the algorithm.
	GCInitWithRect GrabCutMode = 0
	// GCInitWithMask makes the function initialize the state using the provided mask.
	// GCInitWithMask and GCInitWithRect can be combined.
	// Then all the pixels outside of the ROI are automatically initialized with GC_BGD.
	GCInitWithMask = 1
	// GCEval means that the algorithm should just resume.
	GCEval = 2
	// GCEvalFreezeModel means that the algorithm should just run a single iteration of the GrabCut algorithm
	// with the fixed model
	GCEvalFreezeModel = 3
)

// Grabcut runs the GrabCut algorithm.
// The function implements the GrabCut image segmentation algorithm.
// For further details, please see:
// https://docs.opencv.org/master/d7/d1b/group__imgproc__misc.html#ga909c1dda50efcbeaa3ce126be862b37f
//
func GrabCut(img Mat, mask *Mat, r image.Rectangle, bgdModel *Mat, fgdModel *Mat, iterCount int, mode GrabCutMode) {
	cRect := C.struct_Rect{
		x:      C.int(r.Min.X),
		y:      C.int(r.Min.Y),
		width:  C.int(r.Size().X),
		height: C.int(r.Size().Y),
	}

	C.GrabCut(img.p, mask.p, cRect, bgdModel.p, fgdModel.p, C.int(iterCount), C.int(mode))
}

// HoughMode is the type for Hough transform variants.
type HoughMode int

const (
	// HoughStandard is the classical or standard Hough transform.
	HoughStandard HoughMode = 0
	// HoughProbabilistic is the probabilistic Hough transform (more efficient
	// in case if the picture contains a few long linear segments).
	HoughProbabilistic = 1
	// HoughMultiScale is the multi-scale variant of the classical Hough
	// transform.
	HoughMultiScale = 2
	// HoughGradient is basically 21HT, described in: HK Yuen, John Princen,
	// John Illingworth, and Josef Kittler. Comparative study of hough
	// transform methods for circle finding. Image and Vision Computing,
	// 8(1):71–77, 1990.
	HoughGradient = 3
)

// HoughCircles finds circles in a grayscale image using the Hough transform.
// The only "method" currently supported is HoughGradient. If you want to pass
// more parameters, please see `HoughCirclesWithParams`.
//
// For further details, please see:
// https://docs.opencv.org/master/dd/d1a/group__imgproc__feature.html#ga47849c3be0d0406ad3ca45db65a25d2d
//
func HoughCircles(src Mat, circles *Mat, method HoughMode, dp, minDist float64) {
	C.HoughCircles(src.p, circles.p, C.int(method), C.double(dp), C.double(minDist))
}

// HoughCirclesWithParams finds circles in a grayscale image using the Hough
// transform. The only "method" currently supported is HoughGradient.
//
// For further details, please see:
// https://docs.opencv.org/master/dd/d1a/group__imgproc__feature.html#ga47849c3be0d0406ad3ca45db65a25d2d
//
func HoughCirclesWithParams(src Mat, circles *Mat, method HoughMode, dp, minDist, param1, param2 float64, minRadius, maxRadius int) {
	C.HoughCirclesWithParams(src.p, circles.p, C.int(method), C.double(dp), C.double(minDist), C.double(param1), C.double(param2), C.int(minRadius), C.int(maxRadius))
}

// HoughLines implements the standard or standard multi-scale Hough transform
// algorithm for line detection. For a good explanation of Hough transform, see:
// http://homepages.inf.ed.ac.uk/rbf/HIPR2/hough.htm
//
// For further details, please see:
// http://docs.opencv.org/master/dd/d1a/group__imgproc__feature.html#ga46b4e588934f6c8dfd509cc6e0e4545a
//
func HoughLines(src Mat, lines *Mat, rho float32, theta float32, threshold int) {
	C.HoughLines(src.p, lines.p, C.double(rho), C.double(theta), C.int(threshold))
}

// HoughLinesP implements the probabilistic Hough transform
// algorithm for line detection. For a good explanation of Hough transform, see:
// http://homepages.inf.ed.ac.uk/rbf/HIPR2/hough.htm
//
// For further details, please see:
// http://docs.opencv.org/master/dd/d1a/group__imgproc__feature.html#ga8618180a5948286384e3b7ca02f6feeb
//
func HoughLinesP(src Mat, lines *Mat, rho float32, theta float32, threshold int) {
	C.HoughLinesP(src.p, lines.p, C.double(rho), C.double(theta), C.int(threshold))
}
func HoughLinesPWithParams(src Mat, lines *Mat, rho float32, theta float32, threshold int, minLineLength float32, maxLineGap float32) {
	C.HoughLinesPWithParams(src.p, lines.p, C.double(rho), C.double(theta), C.int(threshold), C.double(minLineLength), C.double(maxLineGap))
}

// HoughLinesPointSet implements the Hough transform algorithm for line
// detection on a set of points. For a good explanation of Hough transform, see:
// http://homepages.inf.ed.ac.uk/rbf/HIPR2/hough.htm
//
// For further details, please see:
// https://docs.opencv.org/master/dd/d1a/group__imgproc__feature.html#ga2858ef61b4e47d1919facac2152a160e
//
func HoughLinesPointSet(points Mat, lines *Mat, linesMax int, threshold int,
	minRho float32, maxRho float32, rhoStep float32,
	minTheta float32, maxTheta float32, thetaStep float32) {
	C.HoughLinesPointSet(points.p, lines.p, C.int(linesMax), C.int(threshold),
		C.double(minRho), C.double(maxRho), C.double(rhoStep),
		C.double(minTheta), C.double(maxTheta), C.double(thetaStep))
}

// Integral calculates one or more integral images for the source image.
// For further details, please see:
// https://docs.opencv.org/master/d7/d1b/group__imgproc__misc.html#ga97b87bec26908237e8ba0f6e96d23e28
//
func Integral(src Mat, sum *Mat, sqsum *Mat, tilted *Mat) {
	C.Integral(src.p, sum.p, sqsum.p, tilted.p)
}

// ThresholdType type of threshold operation.
type ThresholdType int

const (
	// ThresholdBinary threshold type
	ThresholdBinary ThresholdType = 0

	// ThresholdBinaryInv threshold type
	ThresholdBinaryInv = 1

	// ThresholdTrunc threshold type
	ThresholdTrunc = 2

	// ThresholdToZero threshold type
	ThresholdToZero = 3

	// ThresholdToZeroInv threshold type
	ThresholdToZeroInv = 4

	// ThresholdMask threshold type
	ThresholdMask = 7

	// ThresholdOtsu threshold type
	ThresholdOtsu = 8

	// ThresholdTriangle threshold type
	ThresholdTriangle = 16
)

// Threshold applies a fixed-level threshold to each array element.
//
// For further details, please see:
// https://docs.opencv.org/3.3.0/d7/d1b/group__imgproc__misc.html#gae8a4a146d1ca78c626a53577199e9c57
//
func Threshold(src Mat, dst *Mat, thresh float32, maxvalue float32, typ ThresholdType) {
	C.Threshold(src.p, dst.p, C.double(thresh), C.double(maxvalue), C.int(typ))
}

// AdaptiveThresholdType type of adaptive threshold operation.
type AdaptiveThresholdType int

const (
	// AdaptiveThresholdMean threshold type
	AdaptiveThresholdMean AdaptiveThresholdType = 0

	// AdaptiveThresholdGaussian threshold type
	AdaptiveThresholdGaussian = 1
)

// AdaptiveThreshold applies a fixed-level threshold to each array element.
//
// For further details, please see:
// https://docs.opencv.org/master/d7/d1b/group__imgproc__misc.html#ga72b913f352e4a1b1b397736707afcde3
//
func AdaptiveThreshold(src Mat, dst *Mat, maxValue float32, adaptiveTyp AdaptiveThresholdType, typ ThresholdType, blockSize int, c float32) {
	C.AdaptiveThreshold(src.p, dst.p, C.double(maxValue), C.int(adaptiveTyp), C.int(typ), C.int(blockSize), C.double(c))
}

// ArrowedLine draws a arrow segment pointing from the first point
// to the second one.
//
// For further details, please see:
// https://docs.opencv.org/master/d6/d6e/group__imgproc__draw.html#ga0a165a3ca093fd488ac709fdf10c05b2
//
func ArrowedLine(img *Mat, pt1 image.Point, pt2 image.Point, c color.RGBA, thickness int) {
	sp1 := C.struct_Point{
		x: C.int(pt1.X),
		y: C.int(pt1.Y),
	}

	sp2 := C.struct_Point{
		x: C.int(pt2.X),
		y: C.int(pt2.Y),
	}

	sColor := C.struct_Scalar{
		val1: C.double(c.B),
		val2: C.double(c.G),
		val3: C.double(c.R),
		val4: C.double(c.A),
	}

	C.ArrowedLine(img.p, sp1, sp2, sColor, C.int(thickness))
}

// Circle draws a circle.
//
// For further details, please see:
// https://docs.opencv.org/master/d6/d6e/group__imgproc__draw.html#gaf10604b069374903dbd0f0488cb43670
//
func Circle(img *Mat, center image.Point, radius int, c color.RGBA, thickness int) {
	pc := C.struct_Point{
		x: C.int(center.X),
		y: C.int(center.Y),
	}

	sColor := C.struct_Scalar{
		val1: C.double(c.B),
		val2: C.double(c.G),
		val3: C.double(c.R),
		val4: C.double(c.A),
	}

	C.Circle(img.p, pc, C.int(radius), sColor, C.int(thickness))
}

// Ellipse draws a simple or thick elliptic arc or fills an ellipse sector.
//
// For further details, please see:
// https://docs.opencv.org/master/d6/d6e/group__imgproc__draw.html#ga28b2267d35786f5f890ca167236cbc69
//
func Ellipse(img *Mat, center, axes image.Point, angle, startAngle, endAngle float64, c color.RGBA, thickness int) {
	pc := C.struct_Point{
		x: C.int(center.X),
		y: C.int(center.Y),
	}
	pa := C.struct_Point{
		x: C.int(axes.X),
		y: C.int(axes.Y),
	}

	sColor := C.struct_Scalar{
		val1: C.double(c.B),
		val2: C.double(c.G),
		val3: C.double(c.R),
		val4: C.double(c.A),
	}

	C.Ellipse(img.p, pc, pa, C.double(angle), C.double(startAngle), C.double(endAngle), sColor, C.int(thickness))
}

// Line draws a line segment connecting two points.
//
// For further details, please see:
// https://docs.opencv.org/master/d6/d6e/group__imgproc__draw.html#ga7078a9fae8c7e7d13d24dac2520ae4a2
//
func Line(img *Mat, pt1 image.Point, pt2 image.Point, c color.RGBA, thickness int) {
	sp1 := C.struct_Point{
		x: C.int(pt1.X),
		y: C.int(pt1.Y),
	}

	sp2 := C.struct_Point{
		x: C.int(pt2.X),
		y: C.int(pt2.Y),
	}

	sColor := C.struct_Scalar{
		val1: C.double(c.B),
		val2: C.double(c.G),
		val3: C.double(c.R),
		val4: C.double(c.A),
	}

	C.Line(img.p, sp1, sp2, sColor, C.int(thickness))
}

// Rectangle draws a simple, thick, or filled up-right rectangle.
// It renders a rectangle with the desired characteristics to the target Mat image.
//
// For further details, please see:
// http://docs.opencv.org/master/d6/d6e/group__imgproc__draw.html#ga346ac30b5c74e9b5137576c9ee9e0e8c
//
func Rectangle(img *Mat, r image.Rectangle, c color.RGBA, thickness int) {
	cRect := C.struct_Rect{
		x:      C.int(r.Min.X),
		y:      C.int(r.Min.Y),
		width:  C.int(r.Size().X),
		height: C.int(r.Size().Y),
	}

	sColor := C.struct_Scalar{
		val1: C.double(c.B),
		val2: C.double(c.G),
		val3: C.double(c.R),
		val4: C.double(c.A),
	}

	C.Rectangle(img.p, cRect, sColor, C.int(thickness))
}

// FillPoly fills the area bounded by one or more polygons.
//
// For more information, see:
// https://docs.opencv.org/master/d6/d6e/group__imgproc__draw.html#gaf30888828337aa4c6b56782b5dfbd4b7
func FillPoly(img *Mat, pts [][]image.Point, c color.RGBA) {
	points := make([]C.struct_Points, len(pts))

	for i, pt := range pts {
		func() {
			p := (*C.struct_Point)(C.malloc(C.size_t(C.sizeof_struct_Point * len(pt))))
			defer C.free(unsafe.Pointer(p))

			pa := getPoints(p, len(pt))

			for j, point := range pt {
				pa[j] = C.struct_Point{
					x: C.int(point.X),
					y: C.int(point.Y),
				}
			}

			points[i] = C.struct_Points{
				points: (*C.Point)(p),
				length: C.int(len(pt)),
			}
		}()
	}

	cPoints := C.struct_Contours{
		contours: (*C.struct_Points)(&points[0]),
		length:   C.int(len(pts)),
	}

	sColor := C.struct_Scalar{
		val1: C.double(c.B),
		val2: C.double(c.G),
		val3: C.double(c.R),
		val4: C.double(c.A),
	}

	C.FillPoly(img.p, cPoints, sColor)
}

// HersheyFont are the font libraries included in OpenCV.
// Only a subset of the available Hershey fonts are supported by OpenCV.
//
// For more information, see:
// http://sources.isc.org/utils/misc/hershey-font.txt
//
type HersheyFont int

const (
	// FontHersheySimplex is normal size sans-serif font.
	FontHersheySimplex HersheyFont = 0
	// FontHersheyPlain issmall size sans-serif font.
	FontHersheyPlain = 1
	// FontHersheyDuplex normal size sans-serif font
	// (more complex than FontHersheySIMPLEX).
	FontHersheyDuplex = 2
	// FontHersheyComplex i a normal size serif font.
	FontHersheyComplex = 3
	// FontHersheyTriplex is a normal size serif font
	// (more complex than FontHersheyCOMPLEX).
	FontHersheyTriplex = 4
	// FontHersheyComplexSmall is a smaller version of FontHersheyCOMPLEX.
	FontHersheyComplexSmall = 5
	// FontHersheyScriptSimplex is a hand-writing style font.
	FontHersheyScriptSimplex = 6
	// FontHersheyScriptComplex is a more complex variant of FontHersheyScriptSimplex.
	FontHersheyScriptComplex = 7
	// FontItalic is the flag for italic font.
	FontItalic = 16
)

// GetTextSize calculates the width and height of a text string.
// It returns an image.Point with the size required to draw text using
// a specific font face, scale, and thickness.
//
// For further details, please see:
// http://docs.opencv.org/master/d6/d6e/group__imgproc__draw.html#ga3d2abfcb995fd2db908c8288199dba82
//
func GetTextSize(text string, fontFace HersheyFont, fontScale float64, thickness int) image.Point {
	cText := C.CString(text)
	defer C.free(unsafe.Pointer(cText))

	sz := C.GetTextSize(cText, C.int(fontFace), C.double(fontScale), C.int(thickness))
	return image.Pt(int(sz.width), int(sz.height))
}

// PutText draws a text string.
// It renders the specified text string into the img Mat at the location
// passed in the "org" param, using the desired font face, font scale,
// color, and line thinkness.
//
// For further details, please see:
// http://docs.opencv.org/master/d6/d6e/group__imgproc__draw.html#ga5126f47f883d730f633d74f07456c576
//
func PutText(img *Mat, text string, org image.Point, fontFace HersheyFont, fontScale float64, c color.RGBA, thickness int) {
	cText := C.CString(text)
	defer C.free(unsafe.Pointer(cText))

	pOrg := C.struct_Point{
		x: C.int(org.X),
		y: C.int(org.Y),
	}

	sColor := C.struct_Scalar{
		val1: C.double(c.B),
		val2: C.double(c.G),
		val3: C.double(c.R),
		val4: C.double(c.A),
	}

	C.PutText(img.p, cText, pOrg, C.int(fontFace), C.double(fontScale), sColor, C.int(thickness))
	return
}

// InterpolationFlags are bit flags that control the interpolation algorithm
// that is used.
type InterpolationFlags int

const (
	// InterpolationNearestNeighbor is nearest neighbor. (fast but low quality)
	InterpolationNearestNeighbor InterpolationFlags = 0

	// InterpolationLinear is bilinear interpolation.
	InterpolationLinear = 1

	// InterpolationCubic is bicube interpolation.
	InterpolationCubic = 2

	// InterpolationArea uses pixel area relation. It is preferred for image
	// decimation as it gives moire-free results.
	InterpolationArea = 3

	// InterpolationLanczos4 is Lanczos interpolation over 8x8 neighborhood.
	InterpolationLanczos4 = 4

	// InterpolationDefault is an alias for InterpolationLinear.
	InterpolationDefault = InterpolationLinear

	// InterpolationMax indicates use maximum interpolation.
	InterpolationMax = 7
)

// Resize resizes an image.
// It resizes the image src down to or up to the specified size, storing the
// result in dst. Note that src and dst may be the same image. If you wish to
// scale by factor, an empty sz may be passed and non-zero fx and fy. Likewise,
// if you wish to scale to an explicit size, a non-empty sz may be passed with
// zero for both fx and fy.
//
// For further details, please see:
// https://docs.opencv.org/master/da/d54/group__imgproc__transform.html#ga47a974309e9102f5f08231edc7e7529d
func Resize(src Mat, dst *Mat, sz image.Point, fx, fy float64, interp InterpolationFlags) {
	pSize := C.struct_Size{
		width:  C.int(sz.X),
		height: C.int(sz.Y),
	}

	C.Resize(src.p, dst.p, pSize, C.double(fx), C.double(fy), C.int(interp))
	return
}

// GetRotationMatrix2D calculates an affine matrix of 2D rotation.
//
// For further details, please see:
// https://docs.opencv.org/master/da/d54/group__imgproc__transform.html#gafbbc470ce83812914a70abfb604f4326
func GetRotationMatrix2D(center image.Point, angle, scale float64) Mat {
	pc := C.struct_Point{
		x: C.int(center.X),
		y: C.int(center.Y),
	}
	return newMat(C.GetRotationMatrix2D(pc, C.double(angle), C.double(scale)))
}

// WarpAffine applies an affine transformation to an image. For more parameters please check WarpAffineWithParams
//
// For further details, please see:
// https://docs.opencv.org/master/da/d54/group__imgproc__transform.html#ga0203d9ee5fcd28d40dbc4a1ea4451983
func WarpAffine(src Mat, dst *Mat, m Mat, sz image.Point) {
	pSize := C.struct_Size{
		width:  C.int(sz.X),
		height: C.int(sz.Y),
	}

	C.WarpAffine(src.p, dst.p, m.p, pSize)
}

// WarpAffineWithParams applies an affine transformation to an image.
//
// For further details, please see:
// https://docs.opencv.org/master/da/d54/group__imgproc__transform.html#ga0203d9ee5fcd28d40dbc4a1ea4451983
func WarpAffineWithParams(src Mat, dst *Mat, m Mat, sz image.Point, flags InterpolationFlags, borderType BorderType, borderValue color.RGBA) {
	pSize := C.struct_Size{
		width:  C.int(sz.X),
		height: C.int(sz.Y),
	}
	bv := C.struct_Scalar{
		val1: C.double(borderValue.B),
		val2: C.double(borderValue.G),
		val3: C.double(borderValue.R),
		val4: C.double(borderValue.A),
	}
	C.WarpAffineWithParams(src.p, dst.p, m.p, pSize, C.int(flags), C.int(borderType), bv)
}

// WarpPerspective applies a perspective transformation to an image.
//
// For further details, please see:
// https://docs.opencv.org/master/da/d54/group__imgproc__transform.html#gaf73673a7e8e18ec6963e3774e6a94b87
func WarpPerspective(src Mat, dst *Mat, m Mat, sz image.Point) {
	pSize := C.struct_Size{
		width:  C.int(sz.X),
		height: C.int(sz.Y),
	}

	C.WarpPerspective(src.p, dst.p, m.p, pSize)
}

// Watershed performs a marker-based image segmentation using the watershed algorithm.
//
// For further details, please see:
// https://docs.opencv.org/master/d7/d1b/group__imgproc__misc.html#ga3267243e4d3f95165d55a618c65ac6e1
func Watershed(image Mat, markers *Mat) {
	C.Watershed(image.p, markers.p)
}

// ColormapTypes are the 12 GNU Octave/MATLAB equivalent colormaps.
//
// For further details, please see:
// https://docs.opencv.org/master/d3/d50/group__imgproc__colormap.html
type ColormapTypes int

// List of the available color maps
//
// For further details, please see:
// https://docs.opencv.org/master/d3/d50/group__imgproc__colormap.html#ga9a805d8262bcbe273f16be9ea2055a65
const (
	ColormapAutumn  ColormapTypes = 0
	ColormapBone                  = 1
	ColormapJet                   = 2
	ColormapWinter                = 3
	ColormapRainbow               = 4
	ColormapOcean                 = 5
	ColormapSummer                = 6
	ColormapSpring                = 7
	ColormapCool                  = 8
	ColormapHsv                   = 9
	ColormapPink                  = 10
	ColormapHot                   = 11
	ColormapParula                = 12
)

// ApplyColorMap applies a GNU Octave/MATLAB equivalent colormap on a given image.
//
// For further details, please see:
// https://docs.opencv.org/master/d3/d50/group__imgproc__colormap.html#gadf478a5e5ff49d8aa24e726ea6f65d15
func ApplyColorMap(src Mat, dst *Mat, colormapType ColormapTypes) {
	C.ApplyColorMap(src.p, dst.p, C.int(colormapType))
}

// ApplyCustomColorMap applies a custom defined colormap on a given image.
//
// For further details, please see:
// https://docs.opencv.org/master/d3/d50/group__imgproc__colormap.html#gacb22288ddccc55f9bd9e6d492b409cae
func ApplyCustomColorMap(src Mat, dst *Mat, customColormap Mat) {
	C.ApplyCustomColorMap(src.p, dst.p, customColormap.p)
}

// GetPerspectiveTransform returns 3x3 perspective transformation for the
// corresponding 4 point pairs.
//
// For further details, please see:
// https://docs.opencv.org/master/da/d54/group__imgproc__transform.html#ga8c1ae0e3589a9d77fffc962c49b22043
func GetPerspectiveTransform(src, dst []image.Point) Mat {
	srcPoints := toCPoints(src)
	dstPoints := toCPoints(dst)
	return newMat(C.GetPerspectiveTransform(srcPoints, dstPoints))
}

// DrawContours draws contours outlines or filled contours.
//
// For further details, please see:
// https://docs.opencv.org/3.3.1/d6/d6e/group__imgproc__draw.html#ga746c0625f1781f1ffc9056259103edbc
func DrawContours(img *Mat, contours [][]image.Point, contourIdx int, c color.RGBA, thickness int) {
	cntrs := make([]C.struct_Points, len(contours))

	for i, contour := range contours {
		func() {
			p := (*C.struct_Point)(C.malloc(C.size_t(C.sizeof_struct_Point * len(contour))))
			defer C.free(unsafe.Pointer(p))

			pa := getPoints(p, len(contour))

			for j, point := range contour {
				pa[j] = C.struct_Point{
					x: C.int(point.X),
					y: C.int(point.Y),
				}
			}

			cntrs[i] = C.struct_Points{
				points: (*C.Point)(p),
				length: C.int(len(contour)),
			}
		}()
	}

	cContours := C.struct_Contours{
		contours: (*C.struct_Points)(&cntrs[0]),
		length:   C.int(len(contours)),
	}

	sColor := C.struct_Scalar{
		val1: C.double(c.B),
		val2: C.double(c.G),
		val3: C.double(c.R),
		val4: C.double(c.A),
	}

	C.DrawContours(img.p, cContours, C.int(contourIdx), sColor, C.int(thickness))
}

// Remap applies a generic geometrical transformation to an image.
//
// For further details, please see:
// https://docs.opencv.org/master/da/d54/group__imgproc__transform.html#gab75ef31ce5cdfb5c44b6da5f3b908ea4
func Remap(src Mat, dst, map1, map2 *Mat, interpolation InterpolationFlags, borderMode BorderType, borderValue color.RGBA) {
	bv := C.struct_Scalar{
		val1: C.double(borderValue.B),
		val2: C.double(borderValue.G),
		val3: C.double(borderValue.R),
		val4: C.double(borderValue.A),
	}
	C.Remap(src.p, dst.p, map1.p, map2.p, C.int(interpolation), C.int(borderMode), bv)
}

// Filter2D applies an arbitrary linear filter to an image.
//
// For further details, please see:
// https://docs.opencv.org/master/d4/d86/group__imgproc__filter.html#ga27c049795ce870216ddfb366086b5a04
func Filter2D(src Mat, dst *Mat, ddepth int, kernel Mat, anchor image.Point, delta float64, borderType BorderType) {
	anchorP := C.struct_Point{
		x: C.int(anchor.X),
		y: C.int(anchor.Y),
	}
	C.Filter2D(src.p, dst.p, C.int(ddepth), kernel.p, anchorP, C.double(delta), C.int(borderType))
}

// SepFilter2D applies a separable linear filter to the image.
//
// For further details, please see:
// https://docs.opencv.org/master/d4/d86/group__imgproc__filter.html#ga910e29ff7d7b105057d1625a4bf6318d
func SepFilter2D(src Mat, dst *Mat, ddepth int, kernelX, kernelY Mat, anchor image.Point, delta float64, borderType BorderType) {
	anchorP := C.struct_Point{
		x: C.int(anchor.X),
		y: C.int(anchor.Y),
	}
	C.SepFilter2D(src.p, dst.p, C.int(ddepth), kernelX.p, kernelY.p, anchorP, C.double(delta), C.int(borderType))
}

// LogPolar remaps an image to semilog-polar coordinates space.
//
// For further details, please see:
// https://docs.opencv.org/master/da/d54/group__imgproc__transform.html#gaec3a0b126a85b5ca2c667b16e0ae022d
func LogPolar(src Mat, dst *Mat, center image.Point, m float64, flags InterpolationFlags) {
	centerP := C.struct_Point{
		x: C.int(center.X),
		y: C.int(center.Y),
	}
	C.LogPolar(src.p, dst.p, centerP, C.double(m), C.int(flags))
}

// DistanceTypes types for Distance Transform and M-estimatorss
//
// For further details, please see:
// https://docs.opencv.org/master/d7/d1b/group__imgproc__misc.html#gaa2bfbebbc5c320526897996aafa1d8eb
type DistanceTypes int

const (
	DistUser   DistanceTypes = 0
	DistL1                   = 1
	DistL2                   = 2
	DistC                    = 3
	DistL12                  = 4
	DistFair                 = 5
	DistWelsch               = 6
	DistHuber                = 7
)

// FitLine fits a line to a 2D or 3D point set.
//
// For further details, please see:
// https://docs.opencv.org/master/d3/dc0/group__imgproc__shape.html#gaf849da1fdafa67ee84b1e9a23b93f91f
func FitLine(pts []image.Point, line *Mat, distType DistanceTypes, param, reps, aeps float64) {
	cPoints := toCPoints(pts)
	C.FitLine(cPoints, line.p, C.int(distType), C.double(param), C.double(reps), C.double(aeps))
}

// CLAHE is a wrapper around the cv::CLAHE algorithm.
type CLAHE struct {
	// C.CLAHE
	p unsafe.Pointer
}

// NewCLAHE returns a new CLAHE algorithm
//
// For further details, please see:
// https://docs.opencv.org/master/d6/db6/classcv_1_1CLAHE.html
//
func NewCLAHE() CLAHE {
	return CLAHE{p: unsafe.Pointer(C.CLAHE_Create())}
}

// NewCLAHEWithParams returns a new CLAHE algorithm
//
// For further details, please see:
// https://docs.opencv.org/master/d6/db6/classcv_1_1CLAHE.html
//
func NewCLAHEWithParams(clipLimit float64, tileGridSize image.Point) CLAHE {
	pSize := C.struct_Size{
		width:  C.int(tileGridSize.X),
		height: C.int(tileGridSize.Y),
	}
	return CLAHE{p: unsafe.Pointer(C.CLAHE_CreateWithParams(C.double(clipLimit), pSize))}
}

// Close CLAHE.
func (c *CLAHE) Close() error {
	C.CLAHE_Close((C.CLAHE)(c.p))
	c.p = nil
	return nil
}

// Apply CLAHE.
//
// For further details, please see:
// https://docs.opencv.org/master/d6/db6/classcv_1_1CLAHE.html#a4e92e0e427de21be8d1fae8dcd862c5e
//
func (c *CLAHE) Apply(src Mat, dst *Mat) {
	C.CLAHE_Apply((C.CLAHE)(c.p), src.p, dst.p)
}