基于背景像素突变检测的交通标志图像分割

为了提高交通标志图像处理过程的效果与效率,根据交通标志图像色彩饱和度空间的灰度直方图中包含的点灰度与区域灰度信息,提出了一种有效确定交通标志图像全局分割阈值的算法。首先分析了基于交通标志图像色彩饱和度空间灰度直方图的一种倒溯标准差的变化规律;然后在此基础上提出了如何选取全局图像分割阈值的方法,并采集了大量交通标志图像进行实验验证,同时,与另外两种在HIS空间下常用的图像分割方法的分割效果进行了对比;最后对算法中的部分参数与实验结果进行了分析,并指出了下一步的研究方向。该算法是一种有效的交通标志图像阈值选取方法,可推广到图像处理其他方面的应用。

Traffic sign image segmentation method based on background pixels mutation analysis

To improve the effect and efficiency of processing traffic sign images, this paper put forward an effective algorithm to obtain the global segmentation threshold based on the histogram information of traffic sign image in color saturation space, the point grey information and region grey information especially. It confirmed a variation of return-standard deviation based on the histogram information of saturation space. And then proposesd a method to obtain the global segmentation threshold. It collected and tested a number of traffic sign images, and made some comparison between other common segmentation methods in HIS space with this algorithm on the basis of the segmentation results of the tested traffic sign images. Finally, it discussed the relationship between some parameters in the algorithm and experimental results, which show that this algorithm is effective in selecting segmentation threshold of traffic sign images and can be generalized to other aspects of image processing fields.