车牌图像预处理研究

Research on image preprocessing of the car plate

为解决由于天气或拍摄角度等原因造成拍摄到的车牌图像模糊、歪斜或缺损的情况,研究了车牌图像预处理技术。对输入的灰度图像进行大小归一化,避免因图像的变形而影响后续的处理。通过灰度拉伸增强图像对比度,通过二值化处理实现图像中背景和对象的分割。采用动态阈值法确定图像二值化的关键阈值,使用带修正的自适应邻域平均法消除图像干扰和噪音,并使用Hough变换和旋转投影相结合的方法实现车牌图像的倾斜校正。实验结果表明,所采用的车牌图像预处理方法对灰度图像可以实现较好的处理效果。

Aiming at sloving the problem that the phenomenon of blur, incline and impairment occurred in the car plate recognition because of weather or shooting angle, the image preprocessing technology was studied. The scale of the grey image for car plate was normalized to avoid distortion. The contrast of the image was enhanced by gray expanding and the key threshold of the binarization algorithm was determined based on the dynamical threshold method. It eliminated interference and noises with adaptive neighborhood average method and realized tilt correction with a new method combining the Hough transform with oriented orthogonal method. The experimental results show that the proposed methods have achieved rather better result.