Hybrid Algorithms for Enhancing Video Quality

Abstract

Digital videos are major part of our everyday life and are used in various applications newlinesuch as military and civilian, surveillance, medical diagnostics, astronomical newlineobservations, interpretation and analysis of biometric data, cinema and entertainment newlineindustry and so on. The reported work in the thesis presents a framework to Adaptive newlineoptimization technique for noise reduction and deblurring. Also, a robust technique for newlineconverting low resolution videos to super resolution (SR) videos is proposed. newlineIt is very difficult to view the video if it is taken in dark light with dark newlinebackground. The contrast of the video must be increased after eliminating various newlinenoises and after removing blurriness in each frame of the video. The noise destroys the newlinevideo structure and reduces video quality. So, Horn Schunck - Laplacian Pyramid newlinetechnique is used to enhance video quality. In this research, an efficient video newlineenhancement technique was implemented with the help of significant optical flow newlineassessment and the Laplacian pyramid technique. However, the functional solutions newlinewere produced and differentiated with the available methods to illustrate the efficiency newlineof the estimated technique. At this point, the Laplacian Pyramid optical flow assessment newlineis carried out with Horn Schunck optical flow assessment which is combined to achieve newlinebetter video quality. Compared to other obtainable techniques, the proposed Horn newlineSchunck-Laplacian Pyramid technology delivered an effective performance by means newlineof PSNR, MSE, and RMSE. newlineExperimental analysis was conducted through CamVid Database. The newlineperformance of the Adaptive F-SCA is evaluated using PSNR, and SSIM. The newlinesuggested technique achieves a maximum PSNR of 29.182 dB and a maximum SSIM newlineof 0.9366, indicating that it is superior to other methods. The proposed hybrid technique newlineoutdoes the prevailing methods with maximum PSNR of 33.5026 dB, maximum SDME newlineof 41.1859 dB and maximum SSIM of 0.6222 respectively. newlineThe future extension of the research will be based on any hybrid optimizations newlinefor the