Genetic programming based texture filtering framework

AbstractA framework is presented, which allows for the automated generation of texturefilters by exploiting the 2D-Lookup algorithm and its optimization by evolution-ary algorithms. To use the framework, the user has to give an original image,containing the structural property-of-interest (e.g. a surface fault), and a binaryimage (goal image), wherein each position of the structural property-of-interestis labeled with the foreground color. Doing so, the framework becomes capableof evolving the configuration of the 2D-Lookup algorithm towards a texture filterfor the structural property-of-interest. Genetic programming (GP) is used asthe evolutionary algorithm. For this GP approach, a filter generator derives twooperations based on formal superoperators from the tree, which represents anindividual of the evolving population. The specification of the 2D-Lookup matrixis performed by a relaxation technique. The approach will be demonstrated ontexture fault examples.Keywords : pattern recognition, scene analysis, image processing, textureanalysis, texture filtering, mathematical morphology, 2D-Lookup algorithm, evo-lutionary algorithms, genetic algorithms, genetic programming, fitness function,relaxation, convolution, ordered weighted averaging, texture numbers, orderedweighted minimum, multilayer backpropagation neural network, crossover,mutation, document preprocessing, visual inspection of surfaces, handwritingextraction, image processing framework