A method of classified RGB fusion of multi-scale curvature attributes based on the image background subtraction
Xingran Fan, Yong Li, Yubang Zhou, Peng Fei He · International Geophysical Conference, Qingdao, China, 17-20 April 2017 · 2017
Summary The phenomenon of information overlapping redundancy widely exists between different scales of curvature attributes. So it is necessary to process curvature fusion for this phenomenon. But the specific information at different scales cannot be reflected in the processing result with traditional PCA method and wavelet image fusion algorithm. In this study, we apply the image background subtraction algorithm to extract characteristic information of each scale curvature attributes (classification treatment) and then these unique information are processed by RGB projection with Gaussian function to obtain the multi-scale curvatures attributes RGB true color grading fusion results. The method was applied to a certain gas shale block (Zhaotong, Sichuan) with actual data analysis. We found that the strata deflection and other structural characteristics in different scales of curvature attributes are well reflected by this method in a same fusion image, which is a good show of different scales of information and response of strength. This method could solve the problem of whitening phenomenon occurred in traditional fusion methods. As for the development fractures of gas shale, the method also performs greatly. This study proposes a new multiscale curvature attributes fusion method and provides a favorable basis for further geological interpretation.