Study Of GIS Vector Edge Update Method Based On Prior Knowledge

Changes of natural environment have profound effect on global resources and environments. Timely and efficient change detection and geographical information update becomes urgent need for the detection techniques of imagery changes in remote sensing and data update in GIS. There still exist defects like low automation and generalization in GIS date update-oriented methods of change detection in remote sensing images. Therefore, methods adaptive for character of data type and practical object are gradually focused on.Aiming at the object edge inside Zhalong Wetland, this paper proposes three prior GIS data and remote sensing images based methods for GIS vector edge update. All of the three methods integrate change detection and update processes. Horizontal-vertical neighboring searching method is implemented based on rough location, convexity-concavity identification and detailed location. The experimental effects reveal closer results to manual method except for human-based threshold value and weak shape-maintenance. To confine regional shape, snake model is introduced to substitute previous vector edge for initial contour and adopt internal energy to enhance shape-maintenance. Combining original snake model and greedy snake model, this paper presents an improved method: grouping snake model which performs grouping snaxels, adaptive stepsize and energetical coefficient self-learning. Experiments proves advantageous over original snake and greedy algorithm snake. Grouping snake alleviates snake's sensitiveness to initial contour location, while offers more desirable shape-maintenance in vision than horizontal-vertical neighboring searching method. Furthermore, to balance evolution and improve accuracy of snake model, previous remote sensing images and GIS vector data are both cited to produce multiple prior knowledge data source, which constructs estimated object contour. Enantiomorphous contour of initial contour is symmetrical with initial contour based on the estimated object contour. Results displays update accuracy gets significantly enhanced. To sum up, by making full use of previous remote sensing images and GIS vector data as prior knowledge, automatic vector edge update method succeeds in integrating change detection and data update process and generating high automation and accuracy.