Skip to Main Content
The edge detection of the images is so important for image processing. There are different methods for improving edge detection. In the present work, a neural networks based system for edge extraction of Chinese winespsila micrographs is presented. Back propagation neural network (BPNN) with an improved activation function employing four adjustable parameters is used. Multi-scale mathematical morphological edge detector is a teacher of the neural network. A bank of Gabor filters at different scales and orientations is used, and the resultant Gabor filtering outputs are employed as the inputs of BPNN.