基于角点检测与颜色连贯性分析的车牌定位算法

摘要

针对低分辨率视频图像,提出一种基于角点检测与颜色连贯性分析的快速车牌定位算法.考虑到车牌具备固定的颜色连贯性特点,首先由车牌背景颜色与车牌字体颜色得出分析掩膜；然后使用较小的阈值和最小角点间距计算得出掩膜内角点；最后对角点进行颜色加权,选择权值最大的分析区域.该方法对图像噪声具有极好的抗干扰性,适用于分辨率较低的视频图像数据.实验表明与其他方法相比,该方法的车牌定位准确率高,平均耗时短.%Considering the problems with regard to low resolution video images, a vehicle license plate locating method based on comer detection and regional colour coherence is put forward. First, taking into acount the characteristic of fixed colour coherence the vehicle license plate has, the analysis mask is obtained from background colour and fonts colour of the license plate, and then the smaller threshold and minimum corner points spacing are employed to calculate comer points within the mask. At last, the corner points are weighted with colour, and the analysis region with largest weights is chosen. This method has excellent performance for image noise in anti-jamming and is suitable for low resolution video image data. Experimental results show that, comparing with other methods, this one locates the vehicle plates with higher accuracy and less time cost.