Research On Recovery Condition And Method Of Compressed Sensing For Remote Sensing Image

As increasing demand for information, in order to avoid signal distortions, the traditional sampling techniques will inevitably lead to massive sampling data, greatly increasing the storage and transmission costs. In addition, the development of remote sensing data, the high-resolution, high spectral and temporal (three high) trend is bound to request the data compression technology, and traditional data compression techniques in the "three high" Application of remote sensing images also exist a number of problems in recognition efficiency and recovery accuracy etc. a lot of literature proposes remote sensing image recognition based on the compressed sensing technology for these problems, and gets a lot of better results. The main works of this paper are:1 Discusses sparse reconstruction models and algorithms based on the wavelet transform, and has been some good sparse reconstruction results.2 Modifies the Sub-threshold Weak Conjugate Gradient Pursuit Method, this method has higher recovery accuracy than the Sub-threshold Orthogonal Matching Pursuit Method and the Sub-threshold Weak Conjugate Gradient Pursuit Method, and conditions in the recovery accuracy allows, by changing the parameters can improve the computing speed of the method. Based on the non-smooth and the sparsity of the compressed sensing reconstruction, this paper discusses trust region method based on non-smooth optimization of compressed sensing reconstruction method.3 Designs and researches measurement matrix based on wavelet transform, and applies to remote sensing image processing, and has some satisfactory results.