Research On Key Algorithms Of Uncooled Infrared Imaging Components Based On FPGA

Uncooled infrared imaging technology is developing with the changes of the times,which makes it gradually play an irreplaceable role in many aspects.The existing infrared imaging system are not up to the standard of the three important indicators for the measurement of the image produced by the infrared imaging system which are image non-uniformity,image noise and image detail due to the manufacturing materials of the infrared detector and the manufacturing process of the detector.Since it is difficult to improve the imaging effect by the manufacturing process of the detector,the only feasible approach to improve the imaging quality is the infrared imaging algorithm.In recent years,FPGA has become the most suitable hardware platform for processing real-time image algorithm because of its advantages in data parallel operation and multistage pipeline.This dissertation carries out research on key algorithms of infrared imaging system with MATLAB as the simulation platform of algorithm for the corresponding design on FPGA.The main contents are as follows:(1)A noise template adaptive non-uniformity correction algorithm is proposed in terms of the existing non-uniformity correction algorithm caused by the fixed non-uniformity of the detector and the non-uniformity caused by the time-related temperature drift noise.We applied classical two-point correction to eliminate the fixed non-uniformity caused by the nature of the detector and the noise template and transverse gradient index to further eliminate the non-uniformity caused by temperature drift noise generated in the long time use of detector.The results showed that the algorithm eliminated the two kinds of non-uniformity of infrared image well.(2)An infrared inter-frame denoising algorithm based on dual determination is proposed to solve the problems of inaccurate determination of pixel gray value changes in the original infrared time denoising algorithm which evaluates the difference between two adjacent frames twice.We applied the number of pixels with consistent gray changes in a pixel field in the first time of evaluation to determine whether they are moving pixels.We determined gray value changes of non-moving pixels through a second judgement of which those exceed the threshold are moving pixels and the rest,are judged to be generated by noise for their gray value changes.Lastly,we assigned,the moving pixels as the value of the new frame,and the remaining pixels as the average value of the corresponding pixels of the two frames.The results showed that the proposed algorithm removed inter-frame noise well.(3)An infrared enhancement algorithm with adaptive allocation of enhancement coefficient was proposed for the problems of detail enhancement mismatch and easy enhancement noise in the original enhancement algorithm.We found the details of the image by guiding filtering of the algorithm,compared its gray value with the average gray value of the whole image,and assigned different enhancement coefficients according to the difference.The results show that the algorithm can reduce noise enhancement when enhancing moderate details.