Finds corner or blob features in an image SYNOPSIS corners = mrgingham.find_points(image) [ corners is an array of shape (N,2), the unordered list of ] [ detected feature coordinates in the image ] This function runs just the initial stage of the mrgingham processing: - pre-process image - find chessboard corners (or blobs) This function does NOT look for the grid in the feature set No broadcasting is supported by this function ARGUMENTS - image: numpy array of shape (H,W) and dtype np.uint8. This is the image we're processing. Must be densely-stored, grayscale image - blobs: optional boolean, defaulting to False. If True, we look for black-on-white blobs, instead of chessboard corners. If blobs: we MUST have image_pyramid_level==0 - image_pyramid_level: optional integer defaulting to -1. This can be given to operate on a downsampled version of the image. 0 means "original image", 1 means "downsample by a factor of 2 in each dimension", 2 means "downsample by a factor of 4 in each dimension" and so on. image_pyramid_level < 0 (the default) means "start with a downsampled image, and then refine the results by repeatedly reducing the downsampling. If blobs: image_pyramid_level==0 is the only allowed option - debug: optional boolean, defaulting to False. If True, this will dump various intermediate results into /tmp and it will report more stuff on standard out. Most of the intermediate results are self-plotting data files. Run them. This does the same thing as "mrgingham --debug"; the Python output is unaffected. RETURNED VALUE A numpy array of shape (N,2) containing unordered pixel coordinates of the discovered features. If no features were found, and array of shape (0,2) is returned.