Python Image Processing Cookbook

Blob detection with LOG scale-space

Perform the following steps for blob detection in the LOG space:

1. Find the blobs in the image using the blob_log() function:

blobs_log = blob_log(np.invert(image), max_sigma=40, num_sigma=10, threshold=.2)

2. Compute radii in the third column:

blobs_log[:, 2] = blobs_log[:, 2] * np.sqrt(2)

3. Loop over the blobs found and draw them on the output image, ignoring blobs with a small radius:

for blob in blobs_log:  y, x, r = blob  if np.pi*r**2 > 150:    c = plt.Circle((x, y), r, color=[0.75] + \        np.random.rand(2).tolist(), linewidth=2, \        fill=True, alpha=.7)    # add the blobs to the plot

If you run the preceding code snippets and plot the images, you will get a diagram along the lines of in the following: