| With the rapid development of remote sensing, high-resolution remote sensing images have been widely used in areas such as information collection, dynamic monitoring and so on, with its advantage of realistic and rapid update. Shadow is one of the basic characteristics of remote sensing images. On one hand, the shadow will cause distortion of the image information, which will have an effect on image segmentation, information extraction and change detection, which is a disadvantage; On the other hand, we can extract the height, shape, relative position and others through shadow, which is an advantage. In terms of this point, there has been a lot of research works to develop algorithms of shadow detection and compensation in recent years. How to detect the shadow quickly and accurately and how to compensate the information covered by the shadow is becoming a hot issue recently. According to literatures and experimental research works, the existing shadow detection and compensation algorithms are either highly targeted, or with flaws, and most of them are mainly stepwise processing procedure. Thus, it is necessary to develop an integrated automatic shadow detection and compensation system. Therefore, this paper intends to design and develop an integrated system for ZY-3, with function of shadow detection and compensation, based on ArcGIS Engine platform, Matlab and GDAL development tools.In this paper, image difference and PCA transform analysis were firstly implemented on the original image, then the multimodal threshold automatic extraction algorithm was used to detect the shaded area. We can get a rough shaded area extracted from the difference image and a relatively accurate shaded area from the PCA transform image. Then a new shaded area could be obtained from a union operation on the above two shaded areas. As we all know, water is very similar to shadow in high-resolution remote sensing image, thus we have to detect water and delete it. If the image contains water, detect water by using multimodal threshold automatic extraction algorithm, and then make a Boolean operator to get the exact shadowed area. Finally, this paper verified and analyzed the reliability of the method. The results showed that the method has good universality, high extraction accuracy and efficiency for ZY-3.Paper takes histogram matching method to compensate the shadow. According to the principle of cumulative histogram matching, firstly conduct the statistical cumulative histogram of shaded areas and its adjacent regions. Then, according to the distribution of the cumulative histogram using table look-up technique to find DN values of them under the same probability. Finally, assigns the DN values of adjacent regions to the shadow areas, makes the probability distribution of brightness in shadow areas as close as possible to the shadow adjacency regions to realize shadow compensation.Shadow detection and compensation involves problems of the uncertainty of remote sensing images, the complexity of algorithm and low degree of automation. In order to improve the efficiency and accuracy, this paper designed an integrated system of shadow detection and compensation for ZY-3, with a combined use of ArcGIS Engine platform, Matlab and GDAL development tools. Among them, key technologies such as data block reads, 2% linear stretch and DLL are used in the shadow detection and compensation implementations. Technical problems such as the read of large amount of image data, the image uncertainty, system scalability and other issues have been solved. This paper implemented the integration and optimization of the system with an improvement of operating efficiency. Experimental results showed that the system has higher precision and efficiency in the shadow detection and compensation of ZY-3, and can be used for batch processing of data. |
PDF Full Text Request