Self-adaptive Canny operator edge detection technique

When edge detection is performed using a Canny algorithm, the gradient image should be processed with 'non-maximum module suppression' and then double thresholds evaluated to detect edges. However, the double thresholds are greatly affected by personal experience. Experiments show that the results of edge detection for different images are obviously different if the identical threshold is employed, which restricted the use of Canny algorithm in practice. To solve this problem, an algorithm is proposed which can adaptively determine the double thresholds based on gradient histogram and minimum interclass variance. With this algorithm, it can self-adaptively calculate the double thresholds for different images without the necessity to setup any parameter artificially. Fuzzy algorithm is adopted to choose edge pixels. Theory and experiments show that the algorithm is effective and correct.