hazy ocean
Hi guys, so I have wrote a program, such that it will read an image into Mat object and then change it into 2D arrays, which then gets constructed back into Mat.
So in the end, the Mat that was reconstructed seems to not able to be read within imread, and instead returns a segmentation fault error. What should I do in this case?
fringe tuskBOT
vagrant bison
Use a debugger to figure out where the error occurred. Trace where the memory you tried to access comes from. Notice that you either didn't allocate any or not enough. Fix it.
hazy ocean
Use a debugger to figure out where the error occurred. Trace where the memory yo...
Program received signal SIGSEGV, Segmentation fault.
0x00007ffff7180a73 in cv::opt_AVX2::cvtScale64f8u(unsigned char const*, unsigned long, unsigned char const*, unsigned long, unsigned char*, unsigned long, cv::Size_<int>, void*) () from /usr/local/lib/libopencv_core.so.407
(gdb) bt
#0 0x00007ffff7180a73 in cv::opt_AVX2::cvtScale64f8u(unsigned char const*, unsigned long, unsigned char const*, unsigned long, unsigned char*, unsigned long, cv::Size_<int>, void*) ()
at /usr/local/lib/libopencv_core.so.407
#1 0x00007ffff6f0629e in cv::Mat::convertTo(cv::_OutputArray const&, int, double, double) const ()
at /usr/local/lib/libopencv_core.so.407
#2 0x00007ffff7f9e9d4 in cvImageWidgetSetImage(_CvImageWidget*, void const*) ()
at /usr/local/lib/libopencv_highgui.so.407
#3 0x00007ffff7fa5e58 in cv::impl::GTKWindow::imshow(cv::_InputArray const&) ()
at /usr/local/lib/libopencv_highgui.so.407
#4 0x00007ffff7f9ab09 in cv::imshow(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, cv::_InputArray const&) () at /usr/local/lib/libopencv_highgui.so.407
#5 0x0000555555556c89 in main(int, char const**) (argc=1, argv=0x7fffffffdc78)
at /home/kenny/01-University/Semester_6/Algorithm_Engineering/project/main.cpp:81
(gdb)
got this from my debugger
seems like its a problem within opencv?
vagrant bison
No, assume that opencv is correct and the error is in your code. It crashed inside of opencv, but was called by main.cpp:81. I'd double-check the arguments in that line.
hazy ocean
its literally ```c++
imshow("Display window", A);
where A is ```c++
Mat A(img.rows, img.cols, CV_64F, flat_img);
That narrows it down to A, whatever that is.
hazy ocean
and flat_img is just the same as the original Mat img, except in a flat array form
what I did was
I made a pipeline of
Mat -> 2D array -> flattened 1D array -> Mat
vagrant bison
Are you sure the size of flat_img is img.rows * img.cols * however many bytes CV_64F uses per pixel?
hazy ocean
first question yes
second one no o_o
wait, what do you mean by that? actually I just jerry rigged the code, and not sure what CV_64F is
vagrant bison
I'm fairly sure I only asked 1 question 🤔
hazy ocean
oh uh
flat_img is indeed img.rows * img.cols
however, im not sure what CV_64F mean
im using a single channel, grey image btw
vagrant bison
wait, what do you mean by that? actually I just jerry rigged the code, and not s...
I don't know either, but I assume it's the image format. RGB, AGBR, ....
hazy ocean
so im guessing its only 8 bits unsigned char(whatever this mean? :o)
since the value is from 0-256
and btw I thought segmentation fault deals only with boundaries?
vagrant bison
CV_64F - 64-bit floating-point numbers
So you assume you need 1 byte per pixel but told it to use 8 byte per pixel which explains why it read past the end.
Maybe try CV_8U or something.
hazy ocean
I see!! very interesting
so its not about the array index itself, but the memory length?
vagrant bison
Well, when opencv indexes array index 1 it assumes it's 64 bit past the pointer.
vagrant bison
So ... difficult to tell the difference between array index and memory length here.
vagrant bison
Looks like a format mismatch.
Maybe the format you read the data into is also not right?
hazy ocean
hmm... shouldn't be I think?
since its just simply a copy of itself
like the original image looks like this
so the whole code can be summarized like this
std::string image_path = samples::findFile("static/input_image_1.jpg");
Mat img = imread(image_path, IMREAD_GRAYSCALE);
// Allocate
int **squares = new int *[img.rows];
for (int x = 0; x < img.rows; ++x)
{
squares[x] = new int[img.cols];
}
// Convert from Mat object to 2D Array
for (int i = 0; i < img.rows; i++)
{
for (int j = 0; j < img.cols; j++)
{
int extracted = img.at<uchar>(i, j);
std::cout << extracted << std::endl;
squares[i][j] = extracted;
}
}
// Convert from 2D Array to Mat object
int *flat_img = new int[img.rows*img.cols];
int k = 0;
for (int i = 0; i < img.rows; i++) {
for (int j = 0; j < img.cols; j++) {
flat_img[k] = squares[i][j];
k++;
}
}
Mat A(img.rows, img.cols, CV_8U, flat_img);
imshow("Display window", img);
and if I replace img with A in imshow
it has the weird format effect thingy
vagrant bison
It's weird that you store unsigned chars in ints. There is also a chance you messed up the order. See if flat_img[k] = squares[j][i]; makes it better.
fringe tuskBOT