Study of spatiotemporal processing algorithms for video and image processing

Resumo

The present paper discusses an important task of noise suppression in images and video encountered in many applications including hydroacoustics. We provide the review of spatiotemporal methods and algorithms of noise suppression. We suggest a new technique for the noise suppression problem based on movement detector and the NLM algorithm (non-local means algorithm). The performance of the algorithms is considered using the filtering procedure of images and video. We also study filtering quality as a function of input parameters and we provide recommendations for selecting input parameters. The technique of estimating the filtering quality is suggested by combining such metrics as peak signal-to-noise ratio and the index of structural similarity.