A Compact Computational Core for Image Processing
A. Paasio, Asko Kananen, Mika Laiho, Kari A. I. Halonen · 2001
In this paper we suggest a new coefficient structure that can be used in many bipolar or gray-scale image processing tasks. In bipolar image processing, almost all the proposed templates in the CNN literature can be executed. In grayscale tasks, a very important subclass, namely the weighted ranked order filtering, can be realized. A compact CNN-UM cell structure is proposed that is based on the new coefficient with additional changes to the CNN-UM core.