Research On Object-oriented High- Resolution SAR Image Change Detection Methods

Change detection is an important technique in environmental earth observation and security, and implies the comparison of remote sensing images from the same regions at different time. Synthetic Aperture Radar (SAR) imagery has been widely used in change detection due to the advantages of all-time and all-weather over optical satellite imagery.The spatial resolution of SAR images is increasing, and it improves the ability to recognize the objects. High resolution SAR image change detection technology has been widely used in many fields, such as agricultural investigation, urban expansion monitoring, natural disaster assessment and so on, this technology has been the important research content of national development planning, which provides guarantee and service for scientific analysis and scientific decision-making.At present, the research methods of SAR image change detection can be divided into two categories: pixel level change detection and object level change detection. The pixel level change detection method is sensitive to the difference between the image and the noise, and the pixel level change detection method separates the existing relationship between the pixels and ignores the integrity of the target. The objectlevel change detection method is based on the object as the basic image processing unit, which can comprehensively utilize the intensity information, spatial information and texture information of the object. In order to improve the detection accuracy of SAR images, this paper focus on the idea of object level, and makes full use of the rich details of SAR image and conducts an in-depth study. The main work and innovation are as follows:1. Summary the research status of change detection and the existing problems, elaborate the SAR image characteristics and commonly used image filtering methods, analysis the advantages and disadvantages of various changes methods, introduce the methods of evaluating the accuracy of SAR image change detection.2. This paper concludes and summarizes the segmentation method of SAR image, and mainly introduces the fractal network evolution segmentation method. In this paper, the non-local three edge filtering method is used before the SAR image segmentation, and the segmentation results are better than the traditional enhanced Frost filter.3. The traditional logarithmic ratio method is based on the single pixel or pixel center neighborhood. In the high resolution SAR image change detection, there is a large number of false alarm detection pixels in pixel level log ratio method, and neighborhood mean log ratio method has a loss of detail information, the edge information is fuzzy. This paper adopts the idea of object-oriented, after the SAR image segmentation vector combination, taking the combined object as the basic processing unit. A change detection algorithm based on the logarithm ratio of the mean value of the graph object is proposed, the algorithm can avoid the speckle noise while preserving the edge information of the change region.4. The mean log ratio method only depends on the gray information of SAR image, when the change information is extracted from the building, because of the shadow of the height of the building in the SAR image, the detection effect is not good. Aiming at this problem, a new method of adaptive change of gray scale and LBP texture histogram based on G statistic distance is proposed. Based on the idea of object-oriented, patch object is the basic processing unit, in order to calculate the gray histogram and LBP texture histogram, the object is the basic processing unit, calculate the histogram distance by G statistics,the G statistic histogram distance and LBP texture histogram G statistic distance differences identified by adaptive weighted graph, CFAR algorithm is adopted to get the two value image threshold segmentation of difference image changes.