针对近地面远距离成像场景下轮廓目标受湍流杂波影响而使匹配误差较大的问题,提出一种基于方向形状上下文和边连续性约束的形状点集匹配识别方法。首先将方向特征嵌入传统的形状上下文来构造一个具有尺度和旋转不变性的特征算子。然后,受模板和目标形状之间的边连续性先验启发,在目标匹配能量代价函数中加入轮廓形状边连续性约束条件以提高形状匹配精度。合成湍流杂波场景和真实远距离成像场景中的形状匹配实验结果表明,和传统方法相比,所提方法能够将杂波场景下的目标匹配误差平均降低约6%,同时还降低了计算复杂度。

Contour targets are affected by turbulence clutters in near-ground remote imaging scenes, leading to large matching errors. To address this problem, we propose a shape point set matching recognition method based on an oriented shape context and an edge continuity constraint. In the proposed method, directional features are embedded into a traditional shape context to construct a feature operator with a scale and rotation invariance. Further, inspired by the priori of edge continuity between the template and target shapes, we add the edge continuity constraint condition of the contour shape into the target matching energy cost function to improve the accuracy of shape matching. The experimental results of shape matching in a synthetic turbulence clutter scene and a real remote imaging scene show that compared with the traditional method, the proposed method can reduce the target matching error by about 6% in clutter scenes and reduce computational complexity.