A Fulfillment Of Infrared Image Detail Enhancement Algorithm Based On CUDA

With the continuous development of infrared imaging technology, the width of image dataappears14,16or more, and the sensitivity is increasing, and the frame rate is also increasing.However as people known, infrared imaging is confronted with concentration of the pixel graylevel and low contrast. To solve these problems, researchers presenting better imageenhancement algorithms, which means that the image enhancement algorithms are morecomplex. At this point there are some contradictions.To solve all the contradictions, we propose CPU+GPU architecture of PC terminator.FPGA and associated circuitry are only collecting the raw image data, and image processing isall remained to PC. CPU is used for resources management and process scheduling, and GPUis used for infrared image processing because it is good at dealing with large-scale data parallelcomputing. So this will maximize the overall operation of PC resources, which can be welladapted to trend of high-speed data rate. This article uses the CUDA software programmingarchitecture solutions provided by NVIDIA. CUDA makes GPU easy to develop, and alsomakes more efficient use of the power of the GPU.Furthermore, on the basis of previous studies, the infrared image detail enhancementtechniques are discussed. Starting from the infrared image feature, simply combing the infraredimage enhancement technology development process, which naturally leads to bilateral filterand its unique advantages. Combined with the advantages of other algorithms, andcharacteristics of the infrared imaging system in this paper, we propose infrared image detailenhancement algorithm based on the bilateral filter.Finally, we achieve a combination of both of the above. Through the infrared system test,we can verify the feasibility and effectiveness of the proposed algorithm. While CPU+GPUarchitecture model is also verified in dealing with complex real-time image processingalgorithms.