针对在摄像机发生抖动时拍摄的红外图像的多点目标关联问题,提出了新的算法,采用了基于特征三角形的多点目标关联算法.根据目标间的相对位置关系,利用特征三角形对相邻两帧的图像进行配准,消除由于背景运动和摄像机抖动引起的目标整体的平移、旋转和缩放.然后根据配准后的结果,采用改进的最近邻法对目标进行关联.试验仿真结果表明:算法精度和稳定性明显高于依靠位置和灰度等信息的传统算法,特别是对于在摄像头抖动情况下拍摄的红外多点目标关联具有较高的正确率.%A new algorithm was proposed aiming at solving problems of infrared multi-point targets association when the image was taken by a shaken camera. An algorithm based on feature-triangle in infrared target tracking was proposed, which made use of the information of the infrared target such as position and gray. The adjacent frame were matched by using the feature-triangle based on the position of the targets to eliminate translation, rotation and zooming of whole targets caused by movement of background and shaking of camera. Then improved nearest neighbor was introduced for target association on the result of image registration. Experimental results show that the proposed algorithm was effective in infrared target tracking, compare with the traditional NN algorithm, the new algorithm can obtain higher accuratcy and improve the stability of the infrared target tracking system, especially with the image that was taken when the camera was shaken.
展开▼
