A novel scheme for optimal edge detection and generic recognition

Expansion matching is a template recognition scheme which optimizes a new similarity measure called Discriminative Signal to Noise Ratio (DSNR) and has been shown to robustly recognize templates under conditions of noise, severe occlusion and superposition. The DSNR criterion is better suited to practical conditions than the traditional SNR since it considers as “noise”, even the off-center response of the filter to the signal itself. We have also recently extended the expansion filter to simultaneously recognize multiple templates while optimizing the DSNR. In this paper we introduce a new family of optimal DSNR edge detectors based on the expansion filter for several edge models. The step edge detector is compared with the widely used Canny Edge Detector (CED). Experimental comparisons show that our edge detector is superior to the CED in terms of DSNR even under very noisy signal conditions. We also use the optimal DSNR multiple template expansion filter for generic recognition of faces. We create a “generic face” filter by requiring the designed filter to elicit equal responses to all the training faces while optimizing DSNR