average pooling C++ error - c++

Question

#include "opencv2/core/core.hpp"
#include "opencv2/imgproc/imgproc.hpp"
#include "opencv2/highgui/highgui.hpp"
#include <math.h>
#include <fstream>
#include <iostream>
using namespace cv;
using namespace std;
#define ATD at<double>
Mat average_pooling2x2(Mat mat, int padding_mathed)
{
int width_remain = mat.cols % 2;
int high_remain = mat.rows % 2;
Mat mat_new;
if (width_remain == 0 && high_remain == 0)
mat.copyTo(mat_new);
else
{
if (padding_mathed == 1)//valid
{
Rect roi = Rect(0, 0, mat.cols - width_remain, mat.rows - high_remain);
mat(roi).copyTo(mat_new);
}
else //same
{
mat.copyTo(mat_new);
if (high_remain != 0)
{
Mat row_add = cv::Mat::zeros(high_remain, mat_new.cols,mat_new.type());
mat_new.push_back(row_add);
}
if (width_remain != 0)
{
Mat col_add = cv::Mat::zeros(width_remain, mat_new.rows, mat_new.type());
mat_new = mat_new.t();
mat_new.push_back(col_add);
mat_new = mat_new.t();
}
}
}
Mat res(mat_new.cols / 2, mat_new.rows / 2, mat_new.type(), Scalar::all(0));
if (mat_new.channels() ==3)
{
for (int i = 0; i < res.rows; i++)//this is where error happened
{
uchar *data_res = res.ptr<uchar>(i);
uchar * data = mat_new.ptr<uchar>(2*i);
uchar * data1 = mat_new.ptr<uchar>(2*i+1);
for (int j = 0; j < res.cols*res.channels(); j = j + 3)
{
data_res[j] = (data[j*2] + data[j*2+3] + data1[j*2] + data1[j*2+3]) / 4;
data_res[j + 1] = (data[j*2+1] + data[j*2+4] + data1[j*2+1] + data1[j*2+4]) / 4;
data_res[j + 2] = (data[j*2+2] + data[j*2+5] + data1[j*2+2] + data1[j*2+5]) / 4;
}
}
}
else
{
for (int i = 0; i<res.rows; i++)
{
for (int j = 0; j<res.cols; j++)
{
Mat temp;
Rect roi = Rect(j * 2, i * 2, 2, 2);
mat_new(roi).copyTo(temp);
double val;
val = sum(temp)[0] / (2 * 2);
res.ATD(i, j) = val;
}
}
}
return res;
}
int main(int argc, char** argv)
{
Mat image = imread("C://Users//Administrator//Desktop//11.jpg");
imshow("???", image);
Mat pooling_image;
average_pooling2x2(image, 2).copyTo(pooling_image);
imshow("???", pooling_image);
waitKey();
return 0;
}
OpenCV Error: Assertion failed (y == 0 || (data && dims >= 1 && (unsigned)y < (unsigned)size.p[0])) in cv::Mat::ptr, file d:\opencv\build\include\opencv2\core\mat.inl.hpp, line 827
reccently I try to implement the average pooling using C++, this is the error when I run the code, it seems that maybe the ptr pointer is out of range. but I just can not figure out where is the problem. Really need some help

Answer 1

If you opened the file that the error message references to, you would see that the ptr() method is defined as follows:
template<typename _Tp> inline _Tp* Mat::ptr(int y)
{
CV_DbgAssert( y == 0 || (data && dims >= 1 && (unsigned)y < (unsigned)size.p[0]) );
return (_Tp*)(data + step.p[0]*y);
}
Everything inside CV_DbgAssert() must evaluate to true - otherwise the program is going to crash at runtime. From that condition, it is clear that you are referring to the row in your program that is outside of Mat boundaries (the variable y above).
In your case, I can see several line where the program is going to crash.
In these lines, the crash happens when i gets equal or greater than res.rows/2 (the first one will crash if res.rows is an odd number):
uchar * data = mat_new.ptr<uchar>(2*i);
uchar * data1 = mat_new.ptr<uchar>(2*i+1);
This loop will also crash, because data_res has only res.cols columns, and you allow j to reach res.cols*res.channels()-1:
for (int j = 0; j < res.cols*res.channels(); j = j + 3)
{
data_res[j] = (data[j*2] + data[j*2+3] + data1[j*2] + data1[j*2+3]) / 4;
data_res[j + 1] = (data[j*2+1] + data[j*2+4] + data1[j*2+1] + data1[j*2+4]) / 4;
data_res[j + 2] = (data[j*2+2] + data[j*2+5] + data1[j*2+2] + data1[j*2+5]) / 4;
}
Also, I believe that here:
Mat res(mat_new.cols / 2, mat_new.rows / 2, mat_new.type(), Scalar::all(0));
you may have accidentaly swapped arguments - res has mat_new.cols/2 rows, whereas I think you wanted it to be mat_new.rows/2.