Real-time Infrared Image Nonuniformity Correction Technology Research

With the development of infrared technology, infrared imaging system has been more widely used. Infrared focal plane array receives attention and rapid development because of its high sensitivity, simple structure, low power consumption and fast response, However, due to the various aspects like the device material and process, infrared focal plane array have nonuniformity, it reduce the temperature resolution of infrared imaging system seriously, affect the quality of imaging, limit the application range of the focal plane array, it is necessary to study the generation of nonuniformity and its correction method. The time-varying spatial fixed pattern noise is the main limiting factor of nonuniformity, the correction methods proposed in this paper are a combination of temporal filtering and spatial filtering.This paper first introduces the development and application of infrared focal plane array devices, and then discuss the causes of nonuniformity on infrared focal plane, and selects the nonuniformity definition and size metrics that is easy to operate, and describes several traditional nonuniformity correction algorithm and its advantages and disadvantages, points out that the adaptive correction method is the inevitable trend of nonuniformity correction method development. Adaptive nonuniformity correction method is described in detail in this paper, the most promising method is temporal high-pass filter correction algorithm, based on the study and compare of the traditional temporal high-pass filter nonuniformity algorithm and its two improved algorithms, proposed an improved adaptive correction algorithm based on bilateral filter, in the improved algorithm, a secondary correction matrix that is derived from the bilateral filter coefficient matrix is introduced, this matrix can make the calculation of the correction parameter more accurate. At last, the multi-scale bilateral filter based on temporal high-pass filter algorithm is presented, this algorithm is based on pyramid decomposition and use different time constant for nonuniformity cumulative in different scales. In the experimental part, by correcting the images sequence with analog noise and the images sequence with actual nonuniformity to prove that the improved correction algorithm based on bilateral filter can achieve a better correction effect than the other two improved algorithms and the multi-scale bilateral filter based on temporal high-pass filter algorithm has a better correction effect than the bilateral filter based on temporal high-pass filter algorithm, it not only can remove the nonuniformity of infrared image effectively, but also has a faster convergence speed.