Image fusion algorithm based on spatial frequency-motivated pulse coupled neural networks in Wavelet based contourlet transform domain

Abstract

This paper proposes a new method for multi-focus image fusion based on peNN (pulse coupled neural networks) and WBeT (Wavelet based contourlet transform). WBeT is associated with peNN and is used in image fusion to make full use of the characteristics of them. Spatial high frequency in WBeT domain is input to motivate peNN. Select high frequency coefficients of the fused image by weighted method of firing times. Experiments are designed to testify the performance of the proposed method. The results show comparing the algorithm of traditional wavelet-based, WBeT-based fusion algorithms, our presented method outperforms existing methods, in both visual effect and objective evaluation criteria, which can exact the edge information of the original images better and can be applied on the detection filed.

A general memristor-based pulse coupled neural network with variable linking coefficient for multi-focus image fusion

“…As a major branch of digital image processing, image fusion is actually a process of integrating complementary information (mainly including the edge and texture information) from all the input images to generate a relatively high-resolution (HR) image which provides a more accurate description of the same scene than any of the individual images [12][13][14][15][16][36][37][38][39][40][41]. In this section, the proposed MPCNN is further applied to the multi-focus image fusion with a more compact network topology.…”

“…The pulse coupled neural network (PCNN) was originally developed by Eckhorn in 1990 based on the experimental observations of synchronous pulse bursts in cat and monkey visual cortex [1,2]. As a biologically inspired neural network model, the PCNN possesses numerous unique properties including pulse coupling, pulse synchronization, multiplication modulation and variable threshold [1][2][3], which makes it an efficient alternative in the field of image processing, such as image enhancement [4,5], image segmentation [6,7], image denoising [8,9], object and edge detection [10,11], image fusion [12][13][14][15][16], and so forth. While the PCNN is definitely a parameter-controlled network system [3], the network parameters estimation issue has been considered as a significant factor affecting the overall performance of all the aforementioned PCNN-based image processing applications.…”