An Iterative Thresholding Segmentation Model Using a Modified Pulse Coupled Neural Network

Authors:
Chao Gao;Dongguo Zhou;Yongcai Guo
Affiliations:
Key Laboratory of Opto-electronic Technology and Systems of the Education Ministry, Chongqing Univeristy, Chongqing, China;Key Laboratory of Opto-electronic Technology and Systems of the Education Ministry, Chongqing Univeristy, Chongqing, China;Key Laboratory of Opto-electronic Technology and Systems of the Education Ministry, Chongqing Univeristy, Chongqing, China
Venue:
Neural Processing Letters
Year:
2014

Citing 11
Cited 0

Level set methods: an overview and some recent results

Journal of Computational Physics
Image thresholding based on the EM algorithm and the generalized Gaussian distribution

Pattern Recognition
Feature linking via synchronization among distributed assemblies: Simulations of results from cat visual cortex

Neural Computation
A Simplified pulse-coupled neural network for adaptive segmentation of fabric defects

Machine Vision and Applications
Pulse-coupled neural networks and one-class support vector machines for geometry invariant texture retrieval

Image and Vision Computing
Letters: Automatic image segmentation based on PCNN with adaptive threshold time constant

Neurocomputing
An adaptive image segmentation method based on a modified pulse coupled neural network

ICNC'06 Proceedings of the Second international conference on Advances in Natural Computation - Volume Part I
Perfect image segmentation using pulse coupled neural networks

IEEE Transactions on Neural Networks
Region growing with pulse-coupled neural networks: an alternative to seeded region growing

IEEE Transactions on Neural Networks
New Spiking Cortical Model for Invariant Texture Retrieval and Image Processing

IEEE Transactions on Neural Networks
A New Automatic Parameter Setting Method of a Simplified PCNN for Image Segmentation

IEEE Transactions on Neural Networks

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper presents a novel iterative thresholding segmentation method based on a modified pulse coupled neural network (PCNN) for partitioning pixels carefully into a corresponding cluster. In the modified model, we initially simplify the two inputs of the original PCNN, and then construct a global neural threshold instead of the original threshold under the specified condition that the neuron will keep on firing once it begins. This threshold is shown to be the cluster center of a region in which corresponding neurons fire, and which can be adaptively updated as soon as neighboring neurons are captured. We then propose a method for automatically adjusting the linking coefficient based on the minimum weighted center distance function. Through iteration, the threshold can be made to converge at the possible real center of object region, thus ensuring that the final result will be obtained automatically. Finally, experiments on several infrared images demonstrate the efficiency of our proposed model. Moreover, based on comparisons with two efficient thresholding methods, a number of PCNN-based models show that our proposed model can segment images with high performance.