OpenCV代码提取： threshold函数的实现

threshold algorithm: The simplest image segmentation method.

All thresholding algorithms take a source image (src) and a threshold value (thresh) as input and produce an output image (dst) by comparing the pixel value at source pixel( x , y ) to the threshold. If src ( x , y ) > thresh , then dst ( x , y ) is assigned a some value. Otherwise dst ( x , y ) is assigned some other value.

Otsu binarization: in simple words, it automatically calculates a threshold value from image histogram for a bimodal image. (For images which are not bimodal,binarization won’t be accurate.). working with bimodal images, Otsu’s algorithmtries to find a threshold value (t) which minimizes the weighted within-class variance. It actually finds a value of t which lies in between two peaks such that variances to both classes are minimum.

Otsu's thresholding method involves iterating through all the possible threshold values and calculating a measure of spread for the pixel levels each side of the threshold, i.e. the pixels that either fall in foreground or background.The aim is to find the threshold value where the sum of foreground and background spreads is at its minimum.

Triangle algorithm: A line is constructed between the maximum of the histogram at brightness bmax and the lowest value bmin in the image. The distance d between the line and the histogram h[b] is computed for all values of b from b = bmin to b = bmax. The brightness value bo where the distance between h[bo] and the line is maximal is the threshold value, that is, threshold = bo. This technique is particularly effective when the object pixels produce a weak peak in the histogram.

图像二值化就是将图像上的像素点的灰度值设置为两个值,一般为0,255或者指定的某个值。

Otsu:

目前fbc_cv库中支持uchar和float两种数据类型，经测试，与OpenCV3.1结果完全一致。

实现代码threshold.hpp：

// fbc_cv is free software and uses the same licence as OpenCV
// Email: fengbingchun@163.com

#ifndef FBC_CV_THRESHOLD_HPP_
#define FBC_CV_THRESHOLD_HPP_

/* reference: include/opencv2/imgproc.hpp
              modules/imgproc/src/thresh.cpp
*/

#include <typeinfo>
#include "core/mat.hpp"
#include "imgproc.hpp"

namespace fbc {

template<typename _Tp, int chs> static double getThreshVal_Otsu_8u(const Mat_<_Tp, chs>& src);
template<typename _Tp, int chs> static double getThreshVal_Triangle_8u(const Mat_<_Tp, chs>& src);
template<typename _Tp, int chs> static void thresh_8u(const Mat_<_Tp, chs>& _src, Mat_<_Tp, chs>& _dst, uchar thresh, uchar maxval, int type);
template<typename _Tp, int chs> static void thresh_32f(const Mat_<_Tp, chs>& _src, Mat_<_Tp, chs>& _dst, float thresh, float maxval, int type);

// applies fixed-level thresholding to a single-channel array
// the Otsu's and Triangle methods are implemented only for 8-bit images
// support type: uchar/float, single-channel
template<typename _Tp, int chs>
double threshold(const Mat_<_Tp, chs>& src, Mat_<_Tp, chs>& dst, double thresh, double maxval, int type)
{
	FBC_Assert(typeid(uchar).name() == typeid(_Tp).name() || typeid(float).name() == typeid(_Tp).name()); // uchar || float
	if (dst.empty()) {
		dst = Mat_<_Tp, chs>(src.rows, src.cols);
	} else {
		FBC_Assert(src.rows == dst.rows && src.cols == dst.cols);
	}

	int automatic_thresh = (type & ~THRESH_MASK);
	type &= THRESH_MASK;

	FBC_Assert(automatic_thresh != (THRESH_OTSU | THRESH_TRIANGLE));
	if (automatic_thresh == THRESH_OTSU) {
		FBC_Assert(sizeof(_Tp) == 1);
		thresh = getThreshVal_Otsu_8u(src);
	} else if (automatic_thresh == THRESH_TRIANGLE) {
		FBC_Assert(sizeof(_Tp) == 1);
		thresh = getThreshVal_Triangle_8u(src);
	}

	if (sizeof(_Tp) == 1) {
		int ithresh = fbcFloor(thresh);
		thresh = ithresh;
		int imaxval = fbcRound(maxval);
		if (type == THRESH_TRUNC)
			imaxval = ithresh;
		imaxval = saturate_cast<uchar>(imaxval);

		if (ithresh < 0 || ithresh >= 255) {
			if (type == THRESH_BINARY || type == THRESH_BINARY_INV ||
				((type == THRESH_TRUNC || type == THRESH_TOZERO_INV) && ithresh < 0) ||
				(type == THRESH_TOZERO && ithresh >= 255)) {
				int v = type == THRESH_BINARY ? (ithresh >= 255 ? 0 : imaxval) :
					type == THRESH_BINARY_INV ? (ithresh >= 255 ? imaxval : 0) :
					/*type == THRESH_TRUNC ? imaxval :*/ 0;
				dst.setTo(v);
			}
			else
				src.copyTo(dst);
			return thresh;
		}
		thresh = ithresh;
		maxval = imaxval;
	} else if (sizeof(_Tp) == 4) {
	} else {
		FBC_Error("UnsupportedFormat");
	}

	if (sizeof(_Tp) == 1) {
		thresh_8u(src, dst, (uchar)thresh, (uchar)maxval, type);
	} else {
		thresh_32f(src, dst, (float)thresh, (float)maxval, type);
	}

	return 0;
}

template<typename _Tp, int chs>
static double getThreshVal_Otsu_8u(const Mat_<_Tp, chs>& _src)
{
	Size size = _src.size();
	const int N = 256;
	int i, j, h[N] = { 0 };

	for (i = 0; i < size.height; i++) {
		const uchar* src = _src.ptr(i);
		j = 0;
		for (; j <= size.width - 4; j += 4) {
			int v0 = src[j], v1 = src[j + 1];
			h[v0]++; h[v1]++;
			v0 = src[j + 2]; v1 = src[j + 3];
			h[v0]++; h[v1]++;
		}
		for (; j < size.width; j++)
			h[src[j]]++;
	}

	double mu = 0, scale = 1. / (size.width*size.height);
	for (i = 0; i < N; i++)
		mu += i*(double)h[i];

	mu *= scale;
	double mu1 = 0, q1 = 0;
	double max_sigma = 0, max_val = 0;

	for (i = 0; i < N; i++) {
		double p_i, q2, mu2, sigma;

		p_i = h[i] * scale;
		mu1 *= q1;
		q1 += p_i;
		q2 = 1. - q1;

		if (std::min(q1, q2) < FLT_EPSILON || std::max(q1, q2) > 1. - FLT_EPSILON)
			continue;

		mu1 = (mu1 + i*p_i) / q1;
		mu2 = (mu - q1*mu1) / q2;
		sigma = q1*q2*(mu1 - mu2)*(mu1 - mu2);
		if (sigma > max_sigma) {
			max_sigma = sigma;
			max_val = i;
		}
	}

	return max_val;
}

template<typename _Tp, int chs>
static double getThreshVal_Triangle_8u(const Mat_<_Tp, chs>& _src)
{
	Size size = _src.size();
	const int N = 256;
	int i, j, h[N] = { 0 };

	for (i = 0; i < size.height; i++) {
		const uchar* src = _src.ptr(i);
		j = 0;
		for (; j <= size.width - 4; j += 4) {
			int v0 = src[j], v1 = src[j + 1];
			h[v0]++; h[v1]++;
			v0 = src[j + 2]; v1 = src[j + 3];
			h[v0]++; h[v1]++;
		}

		for (; j < size.width; j++)
			h[src[j]]++;
	}

	int left_bound = 0, right_bound = 0, max_ind = 0, max = 0;
	int temp;
	bool isflipped = false;

	for (i = 0; i < N; i++) {
		if (h[i] > 0) {
			left_bound = i;
			break;
		}
	}
	if (left_bound > 0)
		left_bound--;

	for (i = N - 1; i > 0; i--) {
		if (h[i] > 0) {
			right_bound = i;
			break;
		}
	}
	if (right_bound < N - 1)
		right_bound++;

	for (i = 0; i < N; i++) {
		if (h[i] > max) {
			max = h[i];
			max_ind = i;
		}
	}

	if (max_ind - left_bound < right_bound - max_ind) {
		isflipped = true;
		i = 0, j = N - 1;
		while (i < j) {
			temp = h[i]; h[i] = h[j]; h[j] = temp;
			i++; j--;
		}
		left_bound = N - 1 - right_bound;
		max_ind = N - 1 - max_ind;
	}

	double thresh = left_bound;
	double a, b, dist = 0, tempdist;

	// We do not need to compute precise distance here. Distance is maximized, so some constants can
	// be omitted. This speeds up a computation a bit.
	a = max; b = left_bound - max_ind;
	for (i = left_bound + 1; i <= max_ind; i++) {
		tempdist = a*i + b*h[i];
		if (tempdist > dist) {
			dist = tempdist;
			thresh = i;
		}
	}
	thresh--;

	if (isflipped)
		thresh = N - 1 - thresh;

	return thresh;
}

template<typename _Tp, int chs>
static void thresh_8u(const Mat_<_Tp, chs>& _src, Mat_<_Tp, chs>& _dst, uchar thresh, uchar maxval, int type)
{
	int i, j, j_scalar = 0;
	uchar tab[256];
	Size roi = _src.size();
	roi.width *= _src.channels;

	switch (type) {
	case THRESH_BINARY:
		for (i = 0; i <= thresh; i++)
			tab[i] = 0;
		for (; i < 256; i++)
			tab[i] = maxval;
		break;
	case THRESH_BINARY_INV:
		for (i = 0; i <= thresh; i++)
			tab[i] = maxval;
		for (; i < 256; i++)
			tab[i] = 0;
		break;
	case THRESH_TRUNC:
		for (i = 0; i <= thresh; i++)
			tab[i] = (uchar)i;
		for (; i < 256; i++)
			tab[i] = thresh;
		break;
	case THRESH_TOZERO:
		for (i = 0; i <= thresh; i++)
			tab[i] = 0;
		for (; i < 256; i++)
			tab[i] = (uchar)i;
		break;
	case THRESH_TOZERO_INV:
		for (i = 0; i <= thresh; i++)
			tab[i] = (uchar)i;
		for (; i < 256; i++)
			tab[i] = 0;
		break;
	default:
		FBC_Error("Unknown threshold type");
	}

	if (j_scalar < roi.width) {
		for (i = 0; i < roi.height; i++) {
			const uchar* src = _src.ptr(i);
			uchar* dst = _dst.ptr(i);
			j = j_scalar;

			for (; j <= roi.width - 4; j += 4) {
				uchar t0 = tab[src[j]];
				uchar t1 = tab[src[j + 1]];

				dst[j] = t0;
				dst[j + 1] = t1;

				t0 = tab[src[j + 2]];
				t1 = tab[src[j + 3]];

				dst[j + 2] = t0;
				dst[j + 3] = t1;
			}

			for (; j < roi.width; j++)
				dst[j] = tab[src[j]];
		}
	}
}

template<typename _Tp, int chs>
static void thresh_32f(const Mat_<_Tp, chs>& _src, Mat_<_Tp, chs>& _dst, float thresh, float maxval, int type)
{
	int i, j;
	Size roi = _src.size();
	roi.width *= _src.channels;
	const float* src = (const float*)_src.ptr();
	float* dst = (float*)_dst.ptr();
	size_t src_step = _src.step / sizeof(src[0]);
	size_t dst_step = _dst.step / sizeof(dst[0]);

	switch (type) {
	case THRESH_BINARY:
		for (i = 0; i < roi.height; i++, src += src_step, dst += dst_step) {
			for (j = 0; j < roi.width; j++)
				dst[j] = src[j] > thresh ? maxval : 0;
		}
		break;

	case THRESH_BINARY_INV:
		for (i = 0; i < roi.height; i++, src += src_step, dst += dst_step) {
			for (j = 0; j < roi.width; j++)
				dst[j] = src[j] <= thresh ? maxval : 0;
		}
		break;

	case THRESH_TRUNC:
		for (i = 0; i < roi.height; i++, src += src_step, dst += dst_step) {
			for (j = 0; j < roi.width; j++)
				dst[j] = std::min(src[j], thresh);
		}
		break;

	case THRESH_TOZERO:
		for (i = 0; i < roi.height; i++, src += src_step, dst += dst_step) {
			for (j = 0; j < roi.width; j++) {
				float v = src[j];
				dst[j] = v > thresh ? v : 0;
			}
		}
		break;

	case THRESH_TOZERO_INV:
		for (i = 0; i < roi.height; i++, src += src_step, dst += dst_step) {
			for (j = 0; j < roi.width; j++) {
				float v = src[j];
				dst[j] = v <= thresh ? v : 0;
			}
		}
		break;
	default:
		FBC_Error("BadArg");
	}
}

} // namespace fbc

#endif // FBC_CV_THRESHOLD_HPP_

测试代码test_threshold.cpp：

#include "test_threshold.hpp"
#include <assert.h>

#include <threshold.hpp>
#include <opencv2/opencv.hpp>

int test_threshold_uchar()
{
	cv::Mat matSrc = cv::imread("E:/GitCode/OpenCV_Test/test_images/lena.png", 1);
	if (!matSrc.data) {
		std::cout << "read image fail" << std::endl;
		return -1;
	}
	cv::cvtColor(matSrc, matSrc, CV_BGR2GRAY);

	int width = matSrc.cols;
	int height = matSrc.rows;
	int types[8] = {0, 1, 2, 3, 4, 7, 8, 16};

	for (int i = 0; i < 8; i++) {
		if (types[i] == 7) continue;
		double thresh = 135.0;
		double maxval = 255.0;

		fbc::Mat_<uchar, 1> mat1(height, width, matSrc.data);
		fbc::Mat_<uchar, 1> mat2(height, width);
		fbc::threshold(mat1, mat2, thresh, maxval, types[i]);

		cv::Mat mat1_(height, width, CV_8UC1, matSrc.data);
		cv::Mat mat2_;
		cv::threshold(mat1_, mat2_, thresh, maxval, types[i]);

		assert(mat2.rows == mat2_.rows && mat2.cols == mat2_.cols && mat2.step == mat2_.step);
		for (int y = 0; y < mat2.rows; y++) {
			const fbc::uchar* p1 = mat2.ptr(y);
			const uchar* p2 = mat2_.ptr(y);

			for (int x = 0; x < mat2.step; x++) {
				assert(p1[x] == p2[x]);
			}
		}
	}

	return 0;
}

int test_threshold_float()
{
	cv::Mat matSrc = cv::imread("E:/GitCode/OpenCV_Test/test_images/lena.png", 1);
	if (!matSrc.data) {
		std::cout << "read image fail" << std::endl;
		return -1;
	}
	cv::cvtColor(matSrc, matSrc, CV_BGR2GRAY);
	matSrc.convertTo(matSrc, CV_32FC1);

	int width = matSrc.cols;
	int height = matSrc.rows;
	int types[6] = { 0, 1, 2, 3, 4, 7 };

	for (int i = 0; i < 6; i++) {
		if (types[i] == 7) continue;
		double thresh = 135.0;
		double maxval = 255.0;

		fbc::Mat_<float, 1> mat1(height, width, matSrc.data);
		fbc::Mat_<float, 1> mat2(height, width);
		fbc::threshold(mat1, mat2, thresh, maxval, types[i]);

		cv::Mat mat1_(height, width, CV_32FC1, matSrc.data);
		cv::Mat mat2_;
		cv::threshold(mat1_, mat2_, thresh, maxval, types[i]);

		assert(mat2.rows == mat2_.rows && mat2.cols == mat2_.cols && mat2.step == mat2_.step);
		for (int y = 0; y < mat2.rows; y++) {
			const fbc::uchar* p1 = mat2.ptr(y);
			const uchar* p2 = mat2_.ptr(y);

			for (int x = 0; x < mat2.step; x++) {
				assert(p1[x] == p2[x]);
			}
		}
	}

	return 0;
}

GitHub： https://github.com/fengbingchun/OpenCV_Test