基于交叉投影熵的车辆目标匹配算法

Target Matching of Vehicle Algorithm Based on Cross-projection Entropy

当前用于复杂交通场景的车辆目标匹配算法主要是基于灰度相关的模板匹配算法,其缺点是受光线变化及噪声干扰影响较大。局部交叉熵算法在匹配精度上有明显提高,但运算量大,难以工程实现。因此,提出将投影变换和局部交叉熵相结合构成特征矢量用于目标匹配,其具有较强的抗光线变化能力,并且在一定的局部遮挡情况下,也具有较好的稳健性,在计算时间上也明显优于局部交叉熵算法。最后,通过实验仿真,验证了算法的有效性。

Algorithms based on gray scale correlation method are frequently used in target matching of vehicle, but they are affected heavily by change of light and noises. Local cross entropy algorithm has higher detection accuracy than gray scale correlation method,but high computational cost. So the i- dea of combining image cross entropy with projection vector is proposed that is introduced to target detection of vehicle in complex traffic scenario. This detection algorithm has good anti-shading capability and good detection result can be obtained in high or low light condition,and also good running time compared with gray scale correlation algorithm. The correctness of simulation result is verified by experiment.