Haze Removal Research For GF-1 Satellite Image

By the development of space technology、computer technology and image processing technology, remote sensing image resolution is becoming higher and higher and coverage is becoming more and more widely. High resolution remote sensing images is becoming more and more widely application in the field of agricultural production, mineral exploration etc. But not all of the remote sensing image is suitable for the practical application, this is because of the influenced by the weather factors, part of the remote sensing image will be affected by the shade of cloud and mist. GF-1 as the first star of high resolution series satellites in China, and it has been used in many areas, the image processing technology have important guiding significance for other satellites, so the single “GF-1” remote sensing image cloud removal has a great practical value. In order to remove the haze in the high resolution satellite images, in this paper, the main research work as follows:1.Analysis and summary the composition of the atmosphere and the definition of the haze and the composition of the cloud, and then introduces the principle of thin cloud imaging and sums up the spatial characteristics, spectrum characteristic of remote sensing images and the relationship between the texture features with the terrain features, according to the principle of thin cloud imaging introduced cloud image degradation model, cloud image signal is main of two parts of the electromagnetic wave signals, one is light cloud reflection of electromagnetic wave, and the other one is ground reflection through clouds, conclusion image processing method and bedding to the back of the experiment.2.Summarizes several suitable principles for single high-resolution image to remove the cloud, including the homomorphic filtering method, wavelet transform method and the new combination of method using the cloud information mainly exist in the low frequency part of the frequency domain for “GF-1” image on cloud image processing,3.Take four kinds of methods to remove the haze, using Matlab tools to deal with the haze for three different resolution "GF-1" satellite images. By setting the poor, mean, standard entropy evaluation parameters to evaluate the processed images, contrast image changes before and after the haze removal. Through a summary of the results of finding out the most suitable method to remove "GF-1" image haze.4.Summarizes the work content for full text, the homomorphic filtering method is too conservative, remove haze effect in general, Wavelet transform method to remove haze effect is good, but the high frequency part of the frequency domain with the clouds information existence. Wavelet transform- homomorphic filtering method has the best effect to remove haze, but will remove part of the background information and feature information. According to the characteristics of high resolution satellite images, Find wavelet transform- homomorphic filtering method to remove the haze deficiencies and put forward the direction of where can be improvement and study.