Modeling Method of Structured Feature Multi-scale Analysis for High Resolution Remote Sensing Image Information Extraction

This paper studies modeling method of multi-scale analysis and information extraction, which based on rich detail information and multi-resolution features of high resolution remote sensing image. Firstly, this dissertation describes a transform domain method for feature decomposition and expression of high resolution remote sensing image, which comprises a multi-scale, multi-channel, and multi-layer transformation feature. Then, subsampling, upsampling, and non-subsampling me-thods are applied to construct the feature structure respectively. Finally, a feature structured multi-scale analysis model is set up. The validity of the feature structured multi-scale analysis model is verified by the modeling process analysis and experimental research of the straight tower model. The result indicates that the proposed method can enhance the flexibility of multi-scale analysis of high resolution image and solve the problem of multi-scale analysis and information extraction effectively.