Reputation: 23
before asking I want to say that I know what is Gabor filter and I used it couple of times.
But I got something I theoretically dont understand. (for example - I got this python + openCV + numpy code)
for filter in filters:
filtered = cv.filter2D(img, cv.CV_8UC3, filter)
filtered = np.reshape(filtered, img.shape)
np.maximum(accumulator, filtered, accumulator)
"filters" - are bank of Gabor filters.
As I know "filtered" is result of convolution in the spatial domain.
np.maximum takes all the values of filtered and compers it with accumulator values and stores only the Max values in accumulator.
What I dont understand and my main question is "What does maximum values of convolution result with gabor filter mean?
Upvotes: 1
Views: 251
Reputation: 2493
Maximum values of convolution with Gabor filters mean that at the spatial point they appear, there is a pattern corresponding to the specific Gabor filter you used. Gabor filters are a Gaussian kernel function modulated by a sinusoidal plane wave, so it means that the corresponding frequency associated with the Gabor filter is present at the specific location where it obtained maximum value.
Upvotes: 1