期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2019
卷号：97
期号：3
页码：740-754
出版社：Journal of Theoretical and Applied
摘要：Pattern matching between two scenes is an important part in image or pattern recognition and registration. It based on finding the best match between two scenes using some selected control points. Image of Solar Planets are somehow difficult in defining distinctive features for matching especially when compared with Earth�s image (ex. Contrast, resolution). Geometrical transformation (scaling and rotation) may affect the registration due to improper selection of correct control points. The degree of distortion due to geometrical transform may affect numbers and accuracy of match. In this article, a new auto-selection method for the reference points is established. First, the scale compensation, a new procedure has evaluated for the range of scaling factor (0.5-1.5). Also, the method was established when there is a rotation effects associated with the scaling one. It was evaluated for rotation range (1o-90o) with the same range of scaling. Second, new matching method for finding the best match between the selected list of control was introduced. This method depends on the distance and angle between corresponding points. This method tested on satellite image of Mars. A computer simulation was build using Matlab Language. The results showed a good performance.
