Research On The Technologies Of Infrared Image Enhancement

In this paper, based on the characteristics of infrared image, we have done asystematic research on the traditional image enhancement method. Some typical imageenhancement theories and algorithms that put forward in recent years are analyzed. Onthe basis of this, we propose an algorithm researched on the detail enhancement andadaptive enhancement in infrared image, and simulate the related algorithms with theinfrared gray images, finally analyze the experiment results.The mainly contents this paper presented can be concluded as follows:(1) Systematic researches have been done for conventional image enhancementtechniques, including some typical methods based on spatial and frequency domain.And some representative image enhancement algorithms proposed in recent years areanalyzed.(2) Researches on the mechanism, the digitalization, the mathematical descriptionand the basic characteristics of infrared image are done.(3) From two aspects including the complexity analysis and the effect of imageenhancement evaluation, the basic theory and method of evaluation system of infraredimage algorithm are described. The general principles of algorithm design aresummarized, and the qualitative analysis and quantitative analysis of two kinds of imageenhancement effect evaluation indicators are given.(4) Because the general linear transformation based algorithm cannot deal with thelarge dynamic background information and the small dynamic details, aimed at imageblurring, low contrast, wide dynamic range and other characteristics of infrared image inlow SNR, two new detail enhancement algorithms of infrared image are proposed interms of the principle of the United States of America FLIR company Digital details(DDE) technology, namely: a DDE algorithm based on hierarchical data processing anddynamic compression of infrared image and an improved DDE algorithm. Then,simulation experiment is done based on a set of measured infrared images and a set ofinfrared images which are extracted from the product manuals of FLIR, and come toperformance analysis of the algorithm. Finally, the effectiveness of the algorithm isverified by the simulation experiment.(5) Since the parameter setting to reunification is difficult in conventionalhistogram equalization (HE) algorithm with its in flexible trans formation and unstableenhancement performance, in order to overcome these defects, we proposed anautomatic equalization and contrast enhancement algorithm for infrared image based onGaussian mixture modeling(GMM). The mixture model is used to simulate the graylevel distribution of image, the corresponding algorithm transforms by the cumulativedistribution function of the leading Gauss component and the input gray level are applied to achieve adaptive equalization and contrast enhancement. The simulationexperiment and performance analysis are done by infrared image with differentresolution, and the effectiveness of the algorithm is verified.The experimental results show that the two kinds of digital detail enhancementbased on the principle of DDE can restrain noises as well as preserve and enhance theimage edges, the outline of the object, the target texture and so on. The adaptiveequalization and contrast enhancement algorithm for infrared image based on Gaussianmixture modeling can greatly improve the quality of original infrared image with theobvious visual improvement effect. When we use the algorithm to equalize the highcontrast of infrared image, the global brightness will be elevated and the high contrast inthe area of edge can be maintained at the same time, therefore it also has a certain effectof detail enhancement.