Research of Images Enhancing Algorithms on Fog or Backlighting Conditions and Implementation with Hardwares
|School||Harbin Institute of Technology|
|Course||Information and Communication Engineering|
|Keywords||Image Processing Image Enhancement Retinex|
Fog or strong backlighting, and other harsh conditions have a significant impact on the field of military reconnaissance and the current video surveillance systems, reduce the quality of the image output, and make us not trust the credibility of the message from the image. In order to ensure that the visual system can work in all weather condition, to improve outdoor surveillance system’s the adaptability of the weather, a study of Image Enhancement algorithm on fog or strong backlighting conditions and implementing with hardware is very necessary. The research can improve the robustness and trustworthiness of surveillance systems. This dissertation mainly focuses on researching the technology of the image enhancement on fog or strong backlighting conditions.In this dissertation, firstly, a variety of traditional algorithms and Retinex algorithms based on the Retina and Cortex Theory are introduced, and then four algorithms which make sense to this issue are analyzed, they are: Histogram Equalization, Homomorphic Filtering, Single-scale Retinex and Multi-scale Retinex. The simulation is finished under the Matlab, by comparing the subjective and objective evaluation of the images which were processed by these four algorithms, the conclusion is: From the quality of processing, Retinex are much better than other traditional algorithms and Multi-scale Retinex are better than Single-scale Retinex but the computation is bigger. It is not good at the implementation with hardware. Considering both the quality of processing and the computation, Single-scale Retinex is chosen.Secondly, because of FPGA can take full advantage of a large number of parallel computing and the use of the advantages of pipeline design, the algorithm can be improved, optimized and implemented with VHDL. After that, it can be simulated under the ModelSim6.1d, the result proved in ModelSim6.1d is almost as same as in Matlab.Finally, the corresponding circuit is designed; the single-scale Retinex algorithm in my own hardware platform is achieved and the image enhancing problem of rapid implementation with hardware under the fog or strong backlighting conditions is solved. It makes a certain sense of reality whether used in the field of military reconnaissance or video surveillance system.