A Contribution to Automatic Design of Image Processing Systems--Breeding Optimized Non-Linear and Oriented Kernels for Texture Analysis

Abstract

The rapid development in image processing technology allows the tackling of application of increasing complexity. For efficient design of application specific systems design automation techniques are required. This paper reports on activities for texture classification employing non-linear oriented kernels configured by evolutionary optimization techniques. Our approach was tested with benchmark and application data from leather inspection and found viable and competitive in both cases.