A method of image recognition using cellular neural networks

摘要

Image recognition in real time with a compact system will have vast applications. Morphological image operations have been developed as a powerful tool to analyze shapes. The computation load of the morphological operations is, however, too heavy for the conventional digital computers to process images in real time. Cellular neural networks (CNNs) are the parallel processing system with analog circuits: CNNs have very high speed processing ability. In this report, a basic method to recognize images is described using morphological operators on discrete-time cellular neural networks (DTCNN). This method can be used for the applications of real time image recognition on DTCNN, although in this report the method have been developed on the simulation software. Some results of character recognitions are presented.