Study On Removing Thin Cloud Of Remote Sensing Image Based On Wavelet Transform

Since 1980 s, new technologies rapidly emerge, achieved progressive development. With the gradual maturity of remote sensing technology, high-resolution remote sensing images are widely used in weather forecasting, urban planning, change monitoring, information collection and so on. However, since the majority of remote sensing satellite images belongs to the resulting optical remote sensing images. Means these images are vulnerable to bad weather. This case caused great disturbance to the effective use of remote sensing in various fields.Clarity of a single scene remote sensing is a key factor in determining the quality of remote sensing images. A higher degree of image clarity equals to higher image effectiveness and utilization. In order to obtain a remote sensing image which is clear and easy to interpret. This paper described the thin cloud imaging model and theories involved. On this basis, several commonly method theories were studied and experimental comparison are used to removing thin cloud from remote sensing images. Including histogram matching method, Laplace image enhancement method, homomorphic filtering, wavelet transform method and so on. Deeply discussion has made to the method of wavelet transform. Combined with the knowledge of image processing, Researching the image model of cloud removed. Improved wavelet transform method to removed Haze from remote sensing image. Finally, use a single scene landsat 8 OLI image as the experimental image to the cloud experiment, From subjective and objective aspects to evaluate the images qualities, proved that the improved method can achieve good results to the removal of cloud.This paper described an improved wavelet transform method by selecting the best wavelet function and wavelet decomposition level to corresponding image, Utilizing high frequency emphasis filtering which in the frequency domain to processing the approximate coefficients of wavelet decomposition. Combined the superiority of two methods to processing image under the time domain and frequency domain. This method can remove the cloud noise and keep the detail information of image to a great extent.