汽车胎号图像的去噪及倾斜调整问题的研究

The research on image denoising and tilt adjusting of car tire

为了实现汽车胎号识别,必须对采集的胎号图像进行一系列预处理。本文着重讨论汽车胎号图像二值化后的部分预处理。采用离散点判断函数或形态学中的开启运算去除离散黑噪声点,运用闭合运算去除白空洞噪声。基于整体图像的倾斜度测量算法调整胎号图像的倾斜。经过一系列预处理,为后续的胎号字符分割及识别打好基础。

In order to recognize car tire , the tire image must be pretreated. In this paper, a part of pre-processions after binarization are discussed. The discrete point determining function or open operation of morphology are used to remove noise and discrete points, and the closed computing is used to remove empty white noise. Based on the algorithm on measuring the tilt of the overall image, the tilt of the whole image is adjusted. A series of pre-treatment provide good base for follow-up segmentation and recognition on tire code image.