No abstract

No abstract

Image denoising is an important method of preprocessing, it is one of the forelands in the field of Computer Graphic and Computer Vision. Astronomical target imaging are most vulnerable to atmospheric turbulence and noise interference, in order to reconstruct the high quality image of the target, we need to restore the high frequency signal of image, but noise also belongs to the high frequency signal, so there will be noise amplification in the reconstruction process. In order to avoid this phenomenon, join image denoising in the process of reconstruction is a feasible solution. This paper mainly research on the principle of four classic denoising algorithm, which are TV, BLS -GSM, NLM and BM3D, we use simulate data for image denoising to analysis the performance of the four algorithms, experiments demonstrate that the four algorithms can remove the noise, the BM3D algorithm not only have high quality of denosing, but also have the highest efficiency at the same time.

No abstract

It is difficult to determine the direction of the motion of the image with less time to analyze the series of images taken in the camera movement at a lower resolution. In this paper, a method of L-K optical flow and S-T corner detection is proposed to study the motion direction of the field of view in the image sequence. On this basis, the weighted least squares method is used to further reduce the error. At the same time, the model is extended from two-dimensional to three-dimensional space, so that the model is more universal.