摘  要：为了实现小型无人机快速自主测距避障，在双目视差测距的基础上，提出了一种机载三目视差测距算法，利用三个视觉传感器测得的数据，推导出简单的距离公式。利用各传感器成像间的相关性，提出了一种加快图像识别方法，通过缩小对图像中障碍物像元的搜索范围，有效地减小了目标搜索运算量，加快了搜索速度，为小型无人机快速自主避障系统的研制创造了条件。   关键词：多目成像；立体视觉；视差测距；图像识别；CCD传感器；区域分割   中图分类号：TP872；TP75        文献标识码：A          文章编号：1672-4984（2008）03-0114-03   A fast parallax distance measuring method for small UAV obstacle avoiding   WANG Yi-fan， CHEN De-rong， ZHANG Li-yan （ School of Aerospace Science and Technology，Beijing University of Technology，Beijing 100081，China）   Abstract: In order to achieve obstacle avoidance of small UAV rapidly and independently by ranging, the paper presents an algorithm that measures distance through three vision sensors on the UAV based on binocular parallax distance measuring. The method used three vision sensors to obtain data, and then calculated some simple distance formulae. Using the correlation of image of vision sensors, an image recognition method which could increase the speed was proposed. It effectively decreased the computation power and improved the processing speed of parallax distance measuring by reducing research range of obstacle pixel in the image. This is a base for realizing autonomous obstacle avoiding system of small UAV.   Key words: Multi-aperture imaging； Stereo vision； Parallax measure distance； Image recognition； CCD； Region segmentation   Editor：liyan