A collection of OpenCV projects I made
To compile these solutions you need to tell the linker where the OpenCV libraries are located, so I uploaded my configuration file opencv_debug.props. Note that I use OpenCV 3.4.0, linked statically
Stuff in here:
- ShapeCV: Heuristic edge-based object detection, no AI/machine learning involved.
How it works: computes the gradient magnitude for every pixel of the image (that is, how 'fast' the color changes just before and after that pixel). The gradient is then thresholded to get a binary image of edges (This is called Canny edge detection), which are then 'read' from the image and stored in arrays of pixels. The relevant ones are drawn in a new image (contours) and then converted to poly lines using the Ramer-Douglas-Peucker algorithm. The various zones delimited by the contours are then flood-filled to highlight the different objects in the image.
