A Modified Pulse Coupled Neural Network Model for Nut Image Segmentation

Abstract

In machine vision, image segmentation is a key step in image processing, but there are problems with nut image segmentation such as their own oil stains, poor antinoise performance, rust pollution, and high noise. To solve these problems, this paper proposes a modified pulse coupled neural network (MPCNN) algorithm. The MPCNN uses a linear modulation model to enhance the feedback incentive effect and speeds up the convergence of the algorithm. Meanwhile, the redesigned the model simplifies the connection between neurons and external inputs. While ignoring its own iteration, the activation function maintains the connection between neighboring neurons and ensures the characteristics of rapidly convergence. Finally, the practical nut images were segmented by the proposed modified algorithm and compared with two existing algorithms. The experimental results showed that MPCNN algorithm for regional consistency evaluation is greater than 0.99, which is of better achievement in image segmentation.