Design And Realize Of Image Noise Reduction Embedded System Based On S Transform

With the rapid development of computer technology, the development of video surveillance based on embedded system industry become a trend.It overcomes the traditional analog CCTV surveillance system transmission distance is limited, cannot be connected to the limits, and do not need to deal with the personal computer analog video signal. Also, based on embedded system video monitoring inherited personal computer video monitoring terminal functional advantages, and overcome the video monitoring system based on personal computer terminal such as voltage coupling the original video signal acquisition, compression, communication is relatively complex, the disadvantage of reliability is not high. When use the camera to obtain images, the light intensity is the main factor lead to generate a lot of noise in the image, especially in the case of night vision video image acquisition, containing a gaussian white noise random distribution. So, first of all on camera images collected noise can improve the recognition rate of the embedded image processing.Based on this purpose, this paper design the ARM and Linux image acquisition and noise reduction processing system.Through the study of traditional image noise reduction method, this paper introduce S transform time-frequency denoising algorithm.For two-dimensional generalized S transform the defect of computational complexity and memory footprint is too large, in this paper, based on one-dimensional generalized S transform time-frequency analysis and two-dimensional generalized S transform time-frequency analysis to improve, through the design image is not at the same time corresponding filtering factor of frequency domain, an improved fast discrete orthogonal generalized S transform algorithm. Using the algorithm of synthetic noise image and actual denoising processing a low illumination image, the results show that the improved fast discrete orthogonal generalized S transform algorithm for gaussian noise image denoising effect is better than that of two-dimensional generalized S transform hard and soft threshold denoising algorithm, and the image noise reduction before the signal noise ratio (SNR) after noise reduction increase to a certain extent, maximize retained the original image information. This system consists of two parts, hardware and software, the hardware part based on ARM9 S3C2440 embedded platform as the core, select embedded Linux operation system as software control system.Based on the above design, realized the image noise reduction based on S transform of embedded system design.