Further results on the DRF method for edge detection

摘要

The magnitude-distance and maximum-value-distance filters areanalyzed, and fast realizations are proposed to implement a DRF cascadefor edge detection. It is shown that the maximum-value-distance filteris well adapted to parallel processing and works two times as fast asthe magnitude-distance filter. It is shown also that the adaptivegradient is an optimal estimate for edge-pixel verification afterzero-crossing detection. The efficiency of the proposed algorithms isconfirmed by experimental results