Research Of IRFPA Non-uniformity Adaptive Correction Based On DSP

Infrared thermal imaging technology has broad prospect in all fields of the national economy, such as military, commercial and civil applications, especially in military fields. However, the non-uniformity of infrared thermal imaging system severely limits the performance and application of imaging system. Therefore, correcting non-uniformity of the infrared focal plane arrays and obtaining high-quality infrared image become one of the most critical image-processing technologies of infrared imaging system.The dissertation studies the non-uniformity adaptive correction of IRFPA using improved temporal high-pass filtering algorithm based on DSP processing platform. Firstly the dissertation explores and studies the current non-uniformity correction approaches of infrared focal plane arrays: reference-source-based algorithm (calibration algorithm) and scene-based non-uniformity correction algorithm. The scene-based correction algorithm can correct the image adaptively with non-uniform changes of IRFPA, so it becomes the research focus and direction.This dissertation uses the scene-based temporal high-pass filtering algorithm to correct non-uniformity combining with the practical applications that require specific condition and the work mode. Firstly temporal high-pass filter is constructed based on human vision neural mechanisms. Then the article analyzes the temporal high-pass filtering algorithm in-depth and the algorithm's frequency and performance. By further analysis, we can adjust the parameter added in the algorithm to better adjust the system cut-off frequency, so we can correct non-uniformity of infrared focal plane arrays better.We build the master-slave system of non-uniformity correction in which PC machine is the host, DES3200-A is the slave machine on hardware implementation of the algorithm. During the emulation process, PC as the main control unit, responsible for transferring the pre-capture video image data to the target board by DSP HPI interface, and also controls the whole system synchronization and displays the corrected image. The DES3200-A target platform on the one hand transmits image using EDMA technology, on the other hand completes temporal high-pass filter correction operation by the core board TMS320C6713 (high-performance floating-point DSP chip). The whole emulation system operation speed increases by using DSP high-speed operation capacity and EDMA technology. TI CCS 2.0 completes whole system work and arithmetic implementation. Experimental results show that temporal high-pass filter algorithm corrects the non-uniformity of infrared focal plane arrays to achieve the desired objective, but the accumulation frame number needs set correctly. The job what said in the dissertation establishes a solid foundation for the realization of engineering application about scene-based correction algorithm for infrared focal plane arrays.