I have reshape a 2d grayscale image to a 1d Mat.I want to access and change each pixel.Each pixel should be 0-255
Mat input_img = imread(argv[1], IMREAD_GRAYSCALE);
Mat input_vector = input_img.reshape(0,1);
std::cout << input_1d_vector.at<int>(0,1);
However it is not ouputing a sensible pixel value.My apologies if its a dumb question , i am new with opencv
so, the correct way to access uchar, single channel image would have been:
mat.at<uchar>(y,x); // NOT int !
but the faster way to do it would be:
uchar *ptr = mat.ptr<uchar>(0); // pointer to row 0
// you can index it like ptr[0] ... ptr[mat.cols-1]
and the reason you see garbage, when printing it out is, that cout treats (u)chars as ascii symbols, so you need to cast it:
cout << (int)value << endl;
but better, print out the whole Mat:
cout << mat << endl;
(no problem with types any more !)
This question was answered before : http://answers.opencv.org/question/19... This shows also a performant way to do it(the thing berak meant with "xy" problem i think).
Asked: 2018-11-09 07:48:12 -0600
Seen: 2,635 times
Last updated: Nov 09 '18
what are you really trying to achieve here ? what is the purpose of it ?
NO, you shouldn't do things per-pixel at all. XY-problem here !
I want to convert the 2d representation ofan image to 1d to apply RLE(run length encoding) on it
different story, then. ;)
is this "homework" ? (smells pretty much like it !)
No its not homework , it just need help in accessing the pixels in the Mat , not with RLE
then again no, you shouldn't.
opencv is a high level matrix library. use appropriate functions, don't reinvent square wheels !