Study Of Bionic False Color Fusion Of Multi-band Images

To facilitate target identification, location and tracking, a perfectly fused false color image should be characterized by prominent targets, vivid details and visually comfortable background color. Successful biotic fusion of such a multi-band image and its evaluation is the purpose of this essay, which is achieved by studying features of the source image, color spaces, human and rattlesnake's visual system.Feature examination of the source image and color spaces helps to find the mapping relations between the various information of the former and the spaces of the latter. On this basis, a simple false color fusion is finished. But its fusion effect is not desirable due to visually uncomfortable background color and indistinct target. Mathematical morphology is then introduced to enhance the target information in the source image and a human- visual-feature-oriented arithmetic, to improve the background color.This essay looks closely into human visual system and its information transmission process, conducts experiments to study the function of the neurons'antagonistic receptive field and thus gets the proper parameter range. This data is used to construct a mathematical model of human visual neuron, which, combined with RGB, HSV and YIQ color space, is applied to form a human visual-feature oriented false color fusion model. The model greatly improves the fusion effect, but fails to create desirable prominence of target and visual comfortability of background color. To overcome this difficulty, the rattlesnake, a species good at processing visible and IR information and its visual features is studied. The research results and OR neuron, AND neuron, enhanced neuron and depressed neuron models are used to imitate six rattlesnake's dual-mode cells'function, which are essential for realizing perfect fusion effect.Evaluation is anther key factor of image fusion study. Any comprehensive evaluation must be made from both subjective and objective perspective. This paper employs statistics and histogram to make subjective evaluation and bases objective evaluation on colorimetry and color spaces'features.