Defect detecting in multi-element photodetector by wavelets and neural networks

We consider the application of wavelet transform and neural networks to solving the problem of defect detection (the lack of elements and the presence of adhesions elements) in multi-element photodetectors by processing their images. It is shown that both methods can be successfully applied to the detection of defects. Found that a method based on wavelet transform requires the manual selection of parameters depending on the size of the processed image. Due to the ability of neural networks to learn, a method for the search for defects with neural networks, automatically adapts to the processed image.