Study On De-noising And Quality Evaluation Of Remote Sensing Image For Pretreatment Based On Wavelet Transform

Remote sensing has been widely used in daily life, such as resource survey, environmental assessment, disaster monitoring and earthquake monitoring. Remote sensing information extraction? remote sensing interpretation and remote sensing system are three important questions in study.And, remote sensing image processing is a key technology for remote sensing information analysis.It is very important in the remote sense application. However, it’s inevitably mixed with a variety of noise to the remote sensing image, which can reduce the utility of remote sensing image. To solve the problem of noise in remote sensing images, a method which is proposed as the core of the noise reduction method based on wavelet transform; In order to test results, it will combine the wavelet transform with grey correlation degree to judge quality of remote sensing image after remote sensing image processing.At first, remote sensing image has been simple data pre-processed by using ERDAS IMAGINE software. Secondly, in order to get the optimal threshold parameters, the adaptive wavelet threshold de-noising algorithm will become the basic method, and the best function will be selected, calculating the optimal wavelet decomposition level and acquiring the optimal threshold parameter by using midpoint method or golden section method. Experiment results show that this method is much better than others. Finally, there are methods——subjective estimate method, frequently-used quantitative detection method (PSNR/MSE) and the wavelet coefficients of gray correlation algorithm——which will be used to judge of remote sensing image quality after pre-treatment. Experiment results show that the wavelet coefficients of gray correlation algorithm is much better and more objective than subjective estimate method and frequently-used quantitative detection method (PSNR/MSE).