Multi-focus image fusion method based on NSCT (Non-Subsampled Contourlet Transform) and depth information incentive PCNN (Pulse Coupled Neural Network)

Abstract

The invention discloses a multi-focus image fusion method based on non-subsampled Contourlet transform and a depth information incentive PCNN (Pulse Coupled Neural Network). The method comprises the following steps of generating a low-frequency sub-band image and a series of high-frequency sub-band images after carrying out multi-scale and multidirectional non-subsampled Contourlet transform on input multi-focus source images; adopting edge information energy based on sub-band coefficients to low-frequency sub-bands to obtain low-frequency sub-band coefficients and adopting a modified PCNN model to high-frequency sub-bands to determine each band-pass sub-band coefficient; and lastly, obtaining a fused image through non-subsampled Contourlet inverse transform. The modified PCNN is mainly embodied in that a factor combined with image depth information through adopting SML (Sum-Modified-Laplacian) capable of describing an image direction and texture information well is taken as input of a model, and most of PCNN-based algorithms take pixel gray values as model input items. The method can be well applied to the field of image fusion, and an experimental result shows that a fusion result which more conforms to an eye vision rule can be provided from both the objective evaluation index and the subjective vision effect.

Description

technical field [0001] The invention belongs to the field of image fusion in image processing, in particular to a multi-focus image fusion method based on non-subsampling Contourlet transform (NSCT) and depth information to stimulate PCNN. Background technique [0002] Multi-focus image fusion technology refers to the fusion of multiple source images whose focus objects are different local objects in the same scene, effectively obtains the clear part of each source image, and finally obtains a high-quality image that contains more comprehensive image information in this space. image. [0003] There are three different levels of image fusion methods: pixel level, feature level, and decision level. The multi-focus image fusion researched by the present invention belongs to the level of pixel-level image fusion, mainly including image fusion based on space domain and image fusion based on transform domain. Most of the current researches are based on transform domain methods. ...

AI Technical Summary

Problems solved by technology

Wavelet transform is the most commonly used transformation domain fusion method. It has good time-frequency local performance and can well preserve image detail information, but it does not have translation invariance and will produce pseudo-Gibbs phenomenon.

In 2002, the Contourlet transform was proposed to solve the problem of fewer subbands in the decomposition direction of the wavelet transform, but it does not have translation invariance, resulting in distortion of the fusion image. Pseudo-Gibbs phenomenon

Images