Segmented phase-only filter binarized with a new error diffusion approach

The use of segmented phase-only filters improves the performance of optical correlators. In contrast to conventional filters that compare the scene image to a single reference, a multidecision correlator undertakes several decisions by integrating information from several references into a single segmented phase-only filter. In practice, the optical implementation of such a filter in a multidecision correlator requires the use of optoelectronic interfaces like binary EASLMs. Despite the very high speed of reconfigurability offered by these elements, binarization limits their performance. By straightforwardly applying the constraint of binarization on the segmented filter, the resulting quantization noise significantly reduces the correlator performance. In the present work a new method for diffusing quantization errors is proposed to optimize the binarization process in a way adapted to our application.