corneliu.cincea's profile - activity

On my machine, the resulting image with that size is a white image. But on the other machine(It has OpenCL available) I

minLoc is a Point I cannot do like you said and can you explain what I should see with this line?

minMaxLoc method from OpenCV300 is returning different values on different machines(Windwos). I am using minMaxLoc metho

This is what i have: rows = 9208 cols = 15152 cv::Mat img_masked = cv::Mat(sourceImage.rows, sourceImage.cols, sourceImage.type()); Error: OpenCV Error: Insufficient memory (Failed to allocate 418558848 bytes) in cv::OutOfMemoryError, file C:\builds\master_PackSlave-win32-vc11-shared\opencv\modules\core\src\alloc.cpp, line 52 OpenCV Error: Assertion failed (u != 0) in cv::Mat::create, file C:\builds\master_PackSlave-win32-vc11-shared\opencv\modules\core\src\matrix.cpp, line 411

Do you know why? Or how I can do differently?

Now I have that exception on this line cv::ocl::setUseOpenCL(false); when i am trying to disable it. First-chance exception at 0x03DE80DD (igdrcl32.dll) in ImageUtils.exe: 0xC0000005: Access violation reading location 0x000008E0. Unhandled exception at 0x03DE80DD (igdrcl32.dll) in ImageUtils.exe: 0xC0000005: Access violation reading location 0x000008E0.

I will try. I attached an image with my matrix which i want to flip.

and when I use bmp images I don't have this exception.

yes I verify for valid data. and I am using release dll for release build

I get a base64 string and I decode it with an alg from internet.

BYTE* imageDecoded; long imageDecodedLength;

if (!BufferFromBase64(image_base64_buff, imageLength, imageDecoded, imageDecodedLength))
{

    return false;
}

First-chance exception at 0x0F5880DD (igdrcl32.dll) in ImageUtils.exe: 0xC0000005: Access violation reading location 0x000008E0. Unhandled exception at 0x0F5880DD (igdrcl32.dll) in ImageUtils.exe: 0xC0000005: Access violation reading location 0x000008E0.

     std::vector<uchar> vectorImageDecoded(imageDecodedLength);
memcpy(&vectorImageDecoded[0], imageDecoded, imageDecodedLength);
cv::Mat matrixDecoded = cv::imdecode(cv::Mat(vectorImageDecoded), cv::IMREAD_UNCHANGED);

cv::Mat res;
cv::flip(matrixDecoded, res, 0);
imageMatrix = res;

Why I get this from that dll? How should I avoid this the image is a .png format

Yep :| thanks.

Thank you :) .

Is not working to link with debug version of opencv_world300(I tried, not working). I am developing a c++ nodejs addon an asynchronous addon and this error happens only if I run multiple calls async. Like I said, randomly it happens. I tried to lock with a mutex the clone and flip method from opencv and is working(I didn't see that crash anymore). Do you know why is happening in this way?

When I am trying to make a clone on a matrix an exception is thrown randomly for clone and for flip too. These exceptions are from opencv_world300.dll.

cv::Mat matrixDecoded = imdecode(cv::Mat(vectorImageDecoded), cv::IMREAD_UNCHANGED); imageMatrix = matrixDecoded.clone();
cv::flip(imageMatrix, imageMatrix, 0);

Somebody knows why is happening? Thank you.

Hi, I isolated the imencode method. I put the method in the constructor of the AsyncWorker derived class and is breaking me the flow with Execute method and OKCallback. But when I put the imencode in OKCallback the flow is working ok for me. So AsyncWorker from NodeJS is behaving weird.

There is possible to create manually the header for bitmap? I am using OpenCV3.0.0 I don't have the class cv::BitmapInfoHeader. I would like to create the header manually and attach to the content.

For small .bmp is working(20x20). I used clone but no difference.
Why is working with .png extension and compression? Is there a method to not use imencode and create manually the bitmap header?

"is not working" - means : That I use the AyncWorker and on Execute method I do my encode above code and HandleOKCallback() is never called so in my javascript I cannot see anything . If I use below snippet is working fine but is a different result that I expect because of compression and .png extension. std::vector<int> params(1); params[0] = CV_IMWRITE_PNG_COMPRESSION; params[1] = 9;

 bool result = cv::imencode(".png", imgMatrix, diffBuffer, params);

This is a C++ Addon in Nodejs and I use AsyncWorker to be async. { ... BYTE* DiffBuff ==> where I keep the image content

cv::Mat imgMatrix = cv::Mat(height, width, CV_8UC3, (void*)DiffBuff);
cv::flip (imgMatrix, imgMatrix, 0);
std::vector<uchar> diffBuffer;

returnValue = EncodeMatrix(imgMatrix, diffBuffer);

BYTE * diffImageBuffer = (BYTE *) malloc(diffBuffer.size());
memcpy(diffImageBuffer, &diffBuffer[0], diffBuffer.size());

BufferToBase64(diffImageBuffer, ...);
...

}

//method which encode bmp bool EncodeMatrix(cv::Mat &imgMatrix, std::vector<uchar> &diffBuffer) { bool result = cv::imencode(".bmp", imgMatrix, diffBuffer, std::vector<int>());

 if(!result)
 {         
     printf(" INVALID_IMENCODE !!! \r\n ");
 }
 return result;

}

I use imencode() method to put the header for the .bmp image. For an image with 20x20 pixels is working but for an image with 400x400 pixels is not working. There is no error when I use large .bmp. The program is a comparer algorithm.

Another question if imencode () is not working with larger images how to create manually the header for the .bmp content?

NOTE: I am using OpenCV 3.0.0