matman's profile - activity

I don't know exactly what you want. If you try to do a NonMaximumSupression try something like (For C++): cv::Mat imgMa

I don't know exactly what you want. If you try to do a NonMaximumSupression try something like (For C++): cv::Mat imgMa

dilate is a maximum filter. Just use it with the desired size and form (MORPH_RECT, MORPH_ELLIPSE, or a custom kernel).

dilate is a maximum filter. Just use it with the desired size and form (MORPH_RECT, MORPH_ELLIPSE, or a custom kernel).

Whats unclear? InputArray and OutputArray are just proxy arguments. You can pass for objectPoints and imagePoints either

Imagine a straight line with one extra pixel on one side. How else should the algorithm detect this single pixel which d

Imagine a straight line with one extra pixel on one side. How else should the algorithm detect this single pixel which d

Imagine a straight line with one extra pixel on one side. How else should the algorithm detect this single pixel which d

In case the contour is not closed (area is 0 or length of contour) you get duplicate contour points and this is the corr

What*s wrong with Image Watch 2017? To watch raw values you just have to zoom in enough. An alternative way would be to

If you use Mat with void* your data alignment is unchanged. This is only a Mat wrapper around your data and does not cop

Take a look here and here

If you deal with contours, take a look at approxPolyDP

How did you measure the performance? For valid results you should run your operation about a hundred times without usin

Simple template matching is not rotation (and scale) invariant. Take a look at this project.

Also this could be interesting link text

I think I remember that a vector wasn't resized when the image is scaled down. So it's size was from calculation before

Add some images and an example code. Also give us some information what system you use and than we can help you ;)

If your convolution is separable you can use external parallelization like in my answer here. When using sepFilter2D you

Your contours are not closed. In this case the contour area is calculated as a line. By the way: the arc length of a 1p

It's the number of stripes you parallel loop is splitted. It is not the number of threads. Here is the implementation fo

It's the number of stripes you parallel loop is splitted. It is not the number of threads. Here is the implementation fo

Correction: You detect everything except the yellow square mask and the distortions. white: foreground black: backgrou

parallel_for_ is a wrapper not only for TBB. It also works with OpenMP, pthreads (Linux), Concurrency and WINRT (Windows

For Canny it is the best using the colour image because the algorithm uses the strongest gradient for each pixel out of

Processing a Sobel filter on a binarized image makes little sense. You can try to smooth the input image a little bit u

Dilate and erode are maximum-/minimum filters. So there is no correct or wrong approach. It simply depends on what you t

This code is currently a mixture of using SSE, Neon and the wrapper interface as well as using functors. I try to give y

This code is currently a mixture of using SSE, Neon and the wrapper interface as well as using functors. I try to give y

Neon seems to be supported (see here). Currently there are some pull requests like this which replaces explicit Neon and

Neon seems to be supported (see here). Currently there are some pull requests like this which replaces explicit Neon and

Currently there is a pull request which replaces explicit Neon and SSE operations with OpenCV's vector wrapper which sav

OpenCV java uses C++ under the hood. There are no java source files for the algorithms itself, but java wrappers. You ca

OpenCV java uses C++ under the hood. There are no java source files for the algorithms itself, but java wrappers. You ca

Take a look at approxPolyDP

On Windows I got ~7,5MB for a typical, small terminal application with static build until OpenCV 3.2. Since OpenCV 3.3 i

Mmh, cv::abs doesn't seem to exist anymore. Even in docs the link is dead. As a workaround you could try: cv::absdiff(s

You can use cv::arrowedLine in OpenCV 3.3 (and 3.2?). Round your points to int and pass it to the function.

Yes it's about it's location. Here is a good explanation. If you are using OpenGL for example you can render in subpixe

In general a list of pixels is a list with brightness values (grayvalues or in case of RGB 3 channels with brightness va

Can you check CV_Assert(srcImage.type() == CV_8UC1); before findContours and are you sure cvtColor isn't throwing tha

Can you check CV_Assert(srcImage.type() == CV_8UC1); before findContours and are you sure cvtColor isn't throwing tha

Your srcImage type is not correct. Make sure it is single channel (which should be) and it's depth is CV_8U.

The source code for mulSpectrums is here

I had never any bigger problems with building OpenCV with contrib modules, but I have to admit that the contrib master b