Coding of Image Data via Correlation Filters for Invariant Pattern Recognition: Some Practical Results

The National Optics Institute is currently carrying out a project in automatic target recognition (ATR) in order to locate, recognize, and track potential targets (e.g. tanks or troop carriers) acquired in real time by an infrared camera. The targets may be camouflaged or par­ tially occluded, in real scenes with high levels of clutters and noise. We present an algorithm based on the Dis­ tance Classifier Correlation Filter (DCCF), a subclass of the Synthetic Discriminant Functions (SDF) family which are often used to solve the ATR problem. We describe the general approach of DCCF-based template matching and the algorithms used, the effect of the train­ ing set on the discrimination performance, the optical implementation, and finally a few practical results. I DCCF algorithm