A spatiotemporal denoising scheme for a mobile device

A fast spatiotemporal denoising algorithm for mobile devices is presented. The noise removal is achieved hierarchically in a multi-resolution framework which is combined with image registration and temporal bilateral filtering. In particular image registration and temporal filtering are designed to reduce computing complexity and memory bandwidth requirement while maintaining the quality of the denoising results. The proposed algorithm is implemented on retail camera products in which a coarse-grained reconfigurable processor is embedded. It is confirmed that the processing time is 2.9 seconds with four 20M pixels images by reducing computing complexity by 11% and the amount of data transfer by 24% comparing to the previous algorithm.