The Feature Extraction And Edge Detection Of The Surface Of The Moon

Moon is one of the nearest nature satellite from the earth, and therefore becomes a primetarget for human space exploration. In recent years, with the development and progress ofscience and technology, exploration of the Moon has become a very hot topic, the research onthe Moon’s image processing technology has become one of the key to achieve lunarexploration. The lunar image goal edge is the most basic feature, which carrys a rich ofimage information, is an important basis on the texture feature extraction, image segmentation,feature extraction and other forms for image analysis, so the moon image edge extractiontechnology has become basic technology for the moon exploration.In this paper, image features of lunar surface and Research made of Edge Detection areintroduced correspondingly. Aimming at the problems of the current image edge detectionalgorithm, new edge detection algorithms are proposed, the main contents as follows:(1) The traditional Canny algorithm filtered by Gaussian function can not take intoaccount the local characteristics information of images, leading to the discontinuous edge ofthe object contour, and impacting test results.And the threshold for edge detection needs to beset in advance, and do not have the adaptive capacity, the improved Canny algorithmproposed in this paper, is the mean filter with a K near instead of Gaussian filter for imagesmoothing, to both remove noises and preserve the gray-scale features of the edge. Whilethe dual threshold selection, non-maxima is used to suppress gradient information of imagesto obtain threshold automatically, and improve self-adaptive algorithm. Finally,theexperiments verify the effectiveness of the new algorithm.(2) Image edge is usually produced in a range of different scales. The traditional singlescale operators of edge detection easily are somewhat missing. And multiscale edge detectioneffectively combined of a plurality of different scale edge, to correctly detect the image edgeof. Therefore this paper makes use of the multiresolution feature of Contourlet transform,putforward the edge detection method that combined the Contourlet transform and Cannyalgorithm. The algorithm firstly decomposes processed image by using of Contourlettransform, and then non-greatly restrain calculate using adaptive Canny algorithm on thedecomposed different scale sub-image respectively. At the same time, distance near singularpoints on the same scale are connected by using a local direction transfrom, and get the edgeof the image. Experimental results show that this algorithm on the processing edge detailand the extracting weak edge has been significantly improved.