Research On The Application Of Rough Adaptive Genetic Algorithm In Image Restoration

As an important branch of digital image processing technology, imagerestoration is considered as a strong support in digital times, and become a hot spotin the field of the digital image. Image restoration is a process，which means torestore the degraded image into original image as close as possible through therecovery model. Image may become vestigial in the process of obtaining or storing.The research on these can be affected on different degree because the relationbetween the image restoration and the aspect of practical application on life andengineering.This paper put forward a kind of rough adaptive genetic algorithm based on thesimple genetic algorithm and applies it to restore motion blurred image andsalt-pepper image. In order to improve the robustness and execution efficiency,following research work has been launched:Firstly, RAGA (Rough Adaptive Genetic Algorithm) comprised of RAM (RoughAdaptive Model) and optimized GA (Genetic Algorithm) was proposed. In thisalgorithm, the model of matching domain, characterization domain and adaptivedomain model together with simple genetic algorithm. Through the principle ofintelligent search on GA, the approximate solution by matching the approximatesolution in the search solution space was processed by different kinds. Meanwhilethe three basic simple genetic operators in the genetic algorithm were optimized toimprove the overall process performance.Secondly, RAGA was used in the recovery operation in motion blurred image.Image restoration model, which applied the genetic algorithm into rough adaptivemodel, needed to be established in this section as for degradation drop qualitativeimage. Specific operation is as follows: Adaptive domain processing was based onadaptive control on the point spread correction scheme, characterization domainmodel on the symmetric domain including three macro symmetric domain processingand one micro symmetrical dimension process, matching domain on the value ofthree types of classification operation according to nearly optimal solution.Experimental results showed that the proposed algorithm can effectively deal withmotion blur, retain degraded images of image edge and enhance its recovery PSNRvalues. Finally, RAGA was applied to deal with the image restoration of salt-peppernoise. In the process of image degradation on salt-pepper, adaptive domain modelrefers to the correcting scheme based on histogram of sub image block. The part ofcharacterization domain model was on the neighbor eight domains. The part ofmatching domain model was the same as motion blurred image restoration.Experimental results showed that RAGA had stronger robustness and betterefficiency.