多特征判断的合作目标识别方法

Cooperative Target Recognition with Multiple Features Judgment

针对视觉导航中合作目标识别问题,提出一种多特征递进判断的快速准确识别算法。通过图像轮廓特征、Hu不变矩特征以及FAST角点特征等多特征,对图像进行层层处理和判断,剔除非目标区域,最终实现合作目标的准确识别。在该过程中对算法进行了一些改进和加速策略,以满足实时性要求。室内外实验表明该算法在不同尺度、形变以及环境干扰等情况下依然可以准确地识别出合作目标,有较高的鲁棒性和适应性。

For solving cooperative target recognition problem in visual navigation, a fast and accurate algorithm based on progressive judgment of multiple features is proposed. The image is processed and judged by multiple features including image contours, Hu moment invariants and FAST corners. The non-target areas are eliminated and the real targets are obtained. In the process, algorithm is improved and some accelerated strategies are adopted to meet real-time requirements. Experiments show that the proposed algorithm can identify cooperative target robustly and adaptively with different distances, different angles and environment disturbance.