论文部分内容阅读
针对序列图像的多车牌定位问题,提出了一种综合车牌特征与梯度分析的定位算法。该算法使用灰度均衡化与中值滤波算法对图像预处理来消除噪声,利用测试样本图像提取出若干有效的车牌参数作为定位系统的输入,实现了全自动定位车牌的目的。并引入MGD梯度分析方法辅助边缘提取算法来提取车牌候选区域,利用HSV颜色模型,使用主颜色分析方法分析候选区域,并通过车牌字符纹理分析方法做进一步的判断。使用Hough轮廓检测方法计算非文本区域块的直线斜率。该算法定位效果较好,鲁棒性强,有很好的工程应用推广价值。
Aiming at the multi-license plate location of sequence images, a positioning algorithm based on the combination of license plate features and gradient analysis is proposed. The algorithm uses the gray-level equalization and median filtering algorithm to pre-process the image to eliminate the noise. Several valid license plate parameters are extracted from the test sample images as the input of the positioning system to achieve the goal of automatic license plate location. The MGD gradient analysis method is introduced to extract the candidate regions of the license plate by using the assistant edge extraction algorithm. The HSV color model is used to analyze the candidate regions by using the master color analysis method, and the further judgment is made by the texture analysis of license plate characters. Use the Hough contour detection method to calculate the slope of the line in non-text region blocks. The algorithm has good positioning effect, strong robustness and good engineering application promotion value.