基于局部熵差的栅格地图拼接方法  

Raster Map Stitching Based on Local Entropy Difference

作　　者：沈建国[1] 戴永刚[1] 

机构地区：[1]华东师范大学电子科学技术系,上海200062

出　　处：《华东师范大学学报（自然科学版）》2002年第4期48-53,共6页Journal of East China Normal University(Natural Science)

摘　　要：栅格地图识别输入系统中往往要对扫描的多块小地图进行镶嵌拼接成为大块完整地图后才作进一步处理。以往的一些图象拼接方法在图象倾斜时不能正确拼接或者有人机交互量过大、处理效率低等缺点,作者提出一种可以自动校正倾斜的拼接方法。方法先定义了局部熵的概念,用熵差匹配的方法寻找到多个匹配点,然后用最小二乘法计算两幅地图的相对倾斜角以及计算出两图的对应坐标关系,经过倾斜校正快速算法处理和"拼缝"的平滑处理后最终实现拼接。搜索匹配中采用序贯检测及跳跃式策略以减少搜索时间。实验结果表明该方法在地图小角度倾斜时能正确拼接,拼接方法效率高,速度快,是栅格地图识别输入系统中地图拼接的一种新方法。In the raster map recognition systems, it needs enchasing or stitching multiple maps into a large integrated map before other process. Some image stitching methods before cann't stitch correctly when the maps are declined to make much man-machine conversation or and lead low efficiency. An automatic method to stitch the declined maps is called as raster map stitching. In the paper the concept of local entropy is defined, and some matching points using entropy-difference are presented and incline-emendation process and sew-smoothing to stitch maps are developed. In the searching process sequence pass-through test and leap-search algebra was adapted to reduce time. The experimental results show that our method can process a certain obliquity, and have high efficiency. It is a novel method for raster map recognition system.

关 键 词：地图拼接 栅格地图识别输入系统 图象拼接 局部熵 序贯检测 倾斜校正 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]