基于分形几何边界提取的图像跟踪方法

Image Tracking Method Based on Fractal-geometry Edge Extraction

为了满足现代武器装备对图像跟踪系统的实时性和跟踪可靠性的双重要求,针对武器装备所跟踪目 标的特点,将基于分形几何的边界提取方法用于图像跟踪系统,提出了一种基于运动区域和边界图像的目标跟踪方 案。文中的边界提取方法不仅能提取出目标的轮廓,还能较好地保留其内部细节,这样通过模板匹配得到的最佳匹 配位置更加可靠。同时,基于运动区域的边界提取,大大减少了计算量,尤其是对于远景的目标跟踪。此外,在利 用边界图像进行匹配的基础上,采用了自适应模板更新策略,目的是尽可能地克服形变、光照等影响,使跟踪过程 准确可靠．仿真实验表明,该方法运算量小,能满足武器装备实时性的要求,并行之有效。

In order to satisfy the double demand of real-time need and tracking reliability of image tracking system Used in modern armament, aiming at the feature of target tracked by armament, edge extraction method was applied based on fractal geometry in image tracking system, and a target tracking scheme on basis of mobile area and edge image was put forward. The edge extraction method not only extracted the outline of the target, but also maintained the inner details better, thus the reliability of the best matching position found by template matching was improved. Meanwhile, edge extraction was just carried on in mobile area to reduce computation greatly, especially for distant tracking： Besides, adaptive template modification tragedy was applied in edge image matching, in order to try the best to overcome the affect of distortion and illumination variation etc and make the tracking process more accurate and reliable. Simulation results demonstrate that this method is simple enough to meet the real-time demand for armament, and is effective.