Technical Studies, Remote Sensing Image Based On The Reciprocal Cell Recovery

Mankind has entered the information age. To a large extent, the result of commercial and military war now, depends on who can share the information power. Therefore, in the world launched a range of new commercial and military transformation. The state, enterprises, organizations are in the active development of Earth observation satellite. At present, the combination of hardware and software to improve the resolution of remote sensing image becomes a new theoretical idea. This paper focused on the use of the above-mentioned mode of thinking to research the ground image restoration methods which can improve the remote sensing image resolution using tilting mode sampling. Thus, In this paper, we factor in remote sensing image degradation as the main line, use the theory of adaptive reciprocal cell, and go into details about the noise, blurring and aliasing within the traditional model and tilting mode of remote sensing image sampling system.What's more, on this basis, we propose the tilting mode sampling restoration framework and the relative restoration methods. Main innovation theory and research results have been proposed as following:(1) Proposed restoration framework which is suitable for tilting mode sampling. First of all, a detailed analysis of 3 elements image degradation model in the remote sensing image acquisition system. Followed by using the theory of adaptive reciprocal cell, we analyse the distribution of noise, blurring and aliasing during the tilting mode sampling remote sensing image acquisition system. Finally, based on the above-mentioned, we proposed the model is composed of sequence from Step 3: (1) generate the adaptive reciprocal cell of tilting mode sampling image acquisition system. (2) Extract the effective spectrum. (3) Defuzzification.(2) A method of non-local mean filter combining the direction information is proposed. It designs a new weight coefficient function by describing the similarity of image that thinking gray-scale information together with direction. In order to obtain accurate image orientation information, in this paper, we adopt mean curvature flow diffusing filter to smooth the structure tensor of noise image. Hence, we get a new nonlinear structure tensor. On this basis, through spectral decomposition to this structure tensor, we could get derection of the image we need. A large number of experiments show that: the new method got better Edge-Preserving and the effect of image denoising.(3) Proposed 2 kinds of tilting mode image defuzzification models, using tilting mode adaptive reciprocal cell. Firstly, the data-fitting term of TV model is rewrite in Fourier domain and defined on the titling mode adaptive reciprocal cell, thus adaptive reciprocal cell based TV regularization model (ARCTV) is constructed. Then, in order to improve the deblurring capacity of the ARCTV model, according to the filter method proposed in chapter 4, we got the non-local regularization model based on the tilting mode adaptive reciprocal cell. Experimental results show: the two methods we proposed in this paper got good results in defuzzification. Between the 2 methods, the second model is better, both in visual effect and SNR values.