Improved PCNN compensation-based AUV image fusion method, processor and system

Abstract

The invention discloses an improved PCNN compensation-based AUV image fusion method, a processor and a system. The method comprises the following steps of subjecting registered AUV source images to NSCT decomposition respectively so as to obtain a corresponding low-frequency sub-band coefficient and a corresponding high-frequency sub-band coefficient; for the low-frequency sub-band coefficient, selecting a low-frequency sub-band fusion coefficient by adopting the rule of improving the spatial frequency; selecting a high-frequency sub-band fusion coefficient for two layers largest in the direction number of the high-frequency sub-band coefficient by adopting a rule largest in absolute value, and selecting a high-frequency sub-band fusion coefficient for all the rest layers by adopting a self-adaptive PCNN model; performing NSCT reconstruction on the selected low-frequency sub-band fusion coefficient and the selected high-frequency sub-band fusion coefficient, and finally obtaining a fused image. The fusion quality of images is effectively improved.

Description

technical field [0001] The invention belongs to the field of image processing, and in particular relates to an AUV image fusion method, processor and system based on improved PCNN compensation. Background technique [0002] In recent years, the role of AUV (autonomous unmanned underwater vehicle) in the fields of ocean exploration, scientific research and military has become more and more prominent, but AUV will be affected by various unfavorable factors in the process of taking pictures near the sea, such as Surge, smog, and night vision, etc., make the quality of the obtained image poor, which is not conducive to subsequent processing and analysis. For the visible light images taken by it, it has the advantages of rich spectral information, high resolution and large dynamic range due to the use of light reflection imaging. , has the ability to penetrate smoke and snow, can work around the clock, and has strong anti-interference ability. The disadvantages are poor imaging ...

AI Technical Summary

Problems solved by technology

However, the directionality of the wavelet transform is limited, the edge and geometric information of the fused image are not fully preserved, and the image contour information is lacking, resulting in a decrease in image quality.

[0005] The Contourlet Transform (CT) method proposed by M.N.Do et al. in 2002 has strong directionality, high resolution, and is more "sparse". However, due to its downsampling operation in the process of decomposition and reconstruction, Lack of translation invariance, resulting in a pseudo-Gibbs phenomenon in the image, which degrades the image quality

The disadvantages of this invention are: the Contourlet transform is carried out in the discrete domain, and the sampling process does not have translation invariance, which will produce pseudo-Gibbs effects and affect the fusion effect; the IPCNN model adopted is complex, has many parameters, and takes a long time to calculate; The rules do not specify further

The disadvantages of this invention are: the PCNN model adopted is complex, has many parameters, and takes a long time to calculate; the explanation is incomplete, and the fusion rules of the sub-band coefficients are not explained; the fusion object is multi-focus images, not different spectral images such as visible light and infrared images.

The disadvantages of this method are: using Laplacian energy (EOL) and visibility (VI) methods to calculate the neuron connection strength value of PCNN, the calculation is complex and time-consuming; the bandpass (high frequency) subband coefficient fusion rule method is single

Images