An Edge Detection Method Based on Optimized BP Neural Network

Abstract

The precision of tile image edge detection has great influence on the dimension detection and defect detection of tile. A parallel model of Back-Propagation (BP) neural network for edge detection of binary image was proposed in this paper, and it was applied to edge detection of gray image. It solved the problem that the convergence speed was too slow to meet the need of training if the BP neural network was used directly to edge detection of gray image because a too huge training sample set was needed. The BP neural network was optimized and solved the problem of unstable detection precision for tile dimension detection. This parallel model was applied to dimension and defect detection of tile, and the precision and speed can meet the requirement of detection precision in tile factories.