An adaptive algorithm of relative radiometric normalization for multi-temporal remote sensing images based on feature
corner was introduced in this paper. The main purpose of this research was to explore an automatic and robust algorithm
of relative radiometric normalization to minimize imaging differences of multi-temporal satellite images. The main idea
was to construct statistical regression model of relative radiometric normalization by extracting steady ground point
correspondences. The algorithm's detailed processes were as follows: First, a method of image matching was applied to
recognize steady ground point correspondences of multi-temporal remote sensing images. Second, a statistical regression
model of relative radiometric normalization was constructed to calibrate imaging difference of multi-temporal remote
sensing images. By the experiments, we could see that the total RSME of reference image and corrected image was
reduced apparently in comparison with the total RSME of reference image and uncalibrated image, and the hue of
corrected image was looked more like that of reference image. We could conclude that this modified algorithm was more
accurate and efficient than traditional algorithm, its adaptive characteristic made it easy to be integrated, and it had more
feasibility and applicable value.
An optimized point matching algorithm is introduced in this paper: the main idea is to extract ground control point by a new feature corner extraction method, then to search the sub-image unequdistantly with dynamic template during template matching calculation. The result of the experiment demonstrates that, the algorithm can extract valuable feature corners, it has more matching accuracy and efficiency, and it has more adaptability and applicable value.
Kinds of historical vector graphs have been gradually accumulated by ground truth data or other reliable sources, but these data have not been fully adopted to detect change in remote sensing circle. In this paper we describe a novel change detection method. The key feature of the new method is the use of a piece of historical land using vector graph. By combing one satellite image and the vector graph after necessary geometric rectification, we could detect change region of the satellite image corresponding to patches in the vector graph. Through adopting coefficient of part change and coefficient of entire change, the study calculates statistics indexes of image corresponding to patches of vector graph with different coefficient groups and assesses the computing results by kappa matrix. According to analytical results, the coefficient of entire change is more important to the number of commission error than the coefficient of part change. This method is benefit to the reuse of historical vector graphs. As the image-processing work of this method is based on patches of historical vector graph, it helps to the development of different vector graphs.