PMMW Image And Visible Image Fusion Algorithm Research Based On An Anti-aliasing Contourlet Transform

Passive millimeter wave (PMMW) imaging is a method of forming images throughthe passive detection of naturally occurring millimeter wave scene radiation. Its uniquetechnical advantages, not only able to image in low visibility conditions, but also cansee concealed weapons under clothing, make it an emerging detecting and imagingsystem. The actual millimeter wave image has low optical resolution, while the visibleimage relatively high. Therefore, the image fusion between the millimeter wave imageand the visible image becomes one of research hot spots in image fusion area.The Contourlet transform, compared to other multiscale multiresolution transforms,can efficiently capture and represent singularities along smooth object boundaries innatural images. Hence, based on a specific research project, this paper researches anddesigns PMMW and visible light image fusion algorithms based on an anti-aliasingContourlet transform. The main contributions of the thesis are shown as below:1. Aimed at the problem that traditional objective evaluation methods neither givehigher importance to visually salient regions in an image, nor take the importance ofedge information into account, we introduce the edge-dependent image fusion qualityindex. Experimental results show that they are in line with HVS, and also have theability to preserve edges and reduce the ringing artifacts around them.2. Aimed at the problem that the basis functions of the original Contourlettransform are bad localized in the frequency domain, leading to frequency aliasing, wedo frequency aliasing graphic analysis, and introduce an anti-aliasing Contourlettransform based on a dual-channel filters structure, and the simulations demonstrate thatthey eliminate the frequency aliasing and the the resulting basis functions are sharplylocalized in the frequency domain.3. Based on the analysis of hazard target characteristics in the millimeter waveimage and background information of the visible image, This paper designs PMMWand visible image fusion algorithms based on an anti-aliasing Contourlet transform, andthen carries out some simulation experiments and true project experiments. Resultsdemonstrate its superiorities in preserving seized hazard target characteristics and removing false targets, compared to original Contourlet transform.4. Against this issue that the fusion rules in image fusion methods have badrobustness, we refer to the edge-dependent image fusion quality index and establish anadaptive fusion rule. We use original Contourlet transform, and anti-aliasing Contourlettransforms having minimum and maximum redundancy to do some simulationexperiments and true project experiments. The results show that backgroundinformation is more precisely reserved in the adaptive fusion rule compared to simplefusion rules.The simulation experiments and true project experiments verify the effectivenessof the above method. The results show that PMMW and visible light image fusionalgorithms based on an anti-aliasing Contourlet transform and an adaptive fusion ruleare very effective in preserving seized hazard target characteristics, removing falsetargets, retaining background information for the benefit of hazard target positioning,.