High-precision Infrared Image Drawing And Evaluation

Because of the passive work, strong anti-interference ability, and all-weatheroperations, infrared imaging technology is widely applied in a variety of civilian andmilitary fields. Since IR imaging possesses specific imaging mechanism, cause theinfrared image generally with the visual effect of fuzzy and contrast is low, thesecharacteristics for the use of infrared image follow-up such as object detection andtracking is extremely unfavorable. Therefore, high-precision infrared image renderingalgorithm research has a very important practical value.In infrared imaging process, first the temperature signal translate into voltagesignal through the temperature sensor, then converted to14-bit or16-bit digital signalby ADC, and finally the data is non-uniformity correction to obtain the original binarydata of the infrared image. This paper studies the14or16-bit image renderingalgorithm for infrared data, and evaluate the quality of the resulting image mappingalgorithm.Firstly, the work principle of infrared thermal imaging and imaging infraredimage characteristic is discussed, and the analysis of advantages and disadvantages ofthe existing image evaluation algorithm, on the basis of the structural similarity imagequality assessment algorithms are not well evaluated a blurred image, designed basedon structural similarity image sharpness gradient without reference to evaluationmethod, evaluation of infrared image clarity, to evaluate the infrared image clarity.Secondly, discussed several infrared image rendering algorithm, summarizes theadvantages and disadvantages of these methods. Based on the frequency of magnitudelarger than the frequency of magnitude smaller amount of information containing more,designed infrared image rendering algorithm probabilistic sequence. Experimentsshow that the algorithm can make the red image details are preserved and enhancedobviously, the whole image rendering is clear, but the algorithm has some limitations.According to the probability sequence algorithm can not adapt to the infrared dataof all types, proposed a transformation method. The infrared data is divided into two sections, using a linear mapping in the background section, delete a lot of redundantinformation, while in the target segment nonlinear transform is used to realize the datamapping, to maximize the retention of the original characteristics of infrared data, andthe data transition as smooth as possible, and not have the too big leap, to maintain theintegrity of the target data. After a comparative analysis of the experiment: thealgorithm improved probabilistic sequence algorithm, so that generating the imagevisual effect is good.