The Remote Sensing Image Retrieval Based On Perception Hash

In recent years, along with the implement of national high-resolution earth observation system. various types of military or civilian satellites was successfully launched, which generated a lot of remote sensing images. How to retrieve useful images from so huge remote sensing image database become an urgent problem. Faced with such a huge amount of data, Retrieve or manually tagging by human is already impossible. If there is a method which can automatically classify remote sensing images and retrieve the interesting ones based on their content information, the speed and efficiency of the image retrieval will be increased dramatically.Image Perception Hash maps the contents of a image to a fixed-length binary code by a hash function. The code is called “image fingerprint” of the image. The perception hash method has robust, compression and other good features, and has been a typical method for network image retrieval. therefore perception hash is an ideal method for remote sensing image retrieval. In this paper, perceptual hashing technology is applied for remote sensing image retrieval. we divide the source image is into blocks, by this way we solve the two remote sensing image retrieval image’s problems,vary large size, and too much redundant information. Based on the characteristics of remote sensing images, we design and implement a four perception hash algorithm applied to remote sensing image retrieval, and established a software platform to test and verify the algorithm’s performance. We analyzed the robust and discriminative of the perception hash algorithms.In order to determine the perception threshold, this paper presents a method for determining the threshold based on the typical target template library, avoiding blindness of the threshold selection process. Finally, we compared the performance of perception hash and traditional Gabor texture features based remote sensing image retrieval method by experiments, the results show that for certain types of remote sensing image,perception hash algorithm can increase retrieval speed dramatically without sacrificing the accuracy of retrieval.