Image Fault Area Detection Algorithm Based on Visual Perception

If the natural scenes decomposed by basic ICA which simulates visual perception then the arrangement in space of its basis functions are in disorder. This result is contradicted with physiological mechanisms of vision. So, a new compute model is proposed to simulate two important mechanisms of vision which are visual cortex receptive field topology construct and synchronous oscillation among neuron group. To solve the problem of train image fault detection, a novel algorithm was proposed based on above compute model. The experiment results show that, the algorithm can increase fault detection rate effectively compared with traditional methods which absence of above two important mechanisms of vision.