A CNN Computing Algorithm for Image Correlation

Abstract

To computing the correlationcoefficients between two images, this paper proposesan algorithm based on the use of cellular neuralnetworks (CNNs), in which most operations(calculations) are achieved by parallel processing.Thus, on the one hand, we can reduce computing time;on the other hand, the computing time will notincrease proportionally with increasing the size of thetemplate images. By integrating the CNN algorithm ona emulated digital CNN-Universal Machineimplemented on FPGA (Field Programmable GateArray) there will be possible to perform some task inreal time in the case of a system developed to assistpeople with visual impairments or in a medicaldiagnosis assistance system, for processing andanalysis of computer tomography images.

Keywords

• cellular neural networks (cnn)
• image correlation
• emulated digital cnn-universal machine
• field programmable gate array (fpga)