一种新的相关跟踪算法的设计与实现

DESIGN AND REALIZATION OF A NEW CORRELATION TRACKER ALGORITHM

提出了一种粗搜索结合精匹配定位的两级图像相关匹配算法 ,该法利用目标中心信息的稳定性 ,并结合图像序列间的相关不变性和目标的自相关性给出了一种新的置信度评估方法 ,在存在复杂背景、类似目标、较强噪声和一定旋转及尺寸变化的情况下仍有较高的匹配精度和跟踪稳定性 ,匹配速度也很快 .还描述了采用 CPL D芯片与 DSP处理器共同实现算法的方法 .

A two stage image matching algorithm was developed, in which coarse searching and accurate location are combined. Taking the advantages of stability of the gray level and other informatin of the object’s center, and the relative invariance of the series images and the self relativity of the object, this method provides a new creditability evaluation method. This new matching algorithm performs perfectly in the condition of complicated background and false objects and high noise, certain rotation and size variance, and is very fast at the same time. Finally, this paper described the realization method of the matching algorithm by use of CPLD and DSP.