传统的基于阈值法的汽车牌照自动识别技术在图片模糊程度超过轻微时,识别率急剧下降。在普通情况下,图片模糊程度主要集中在轻微-中度之间,如何解决这一部分图片的识别问题很有意义。本文根据车牌识别过程中,运动造成图像模糊的特点,分析了匀速直线运动时模糊图像的退化模型和恢复模型。提出了适合本系统环境下的点扩散函数模式,结合维纳滤波的方法复原图像。充分利用维纳滤波在处理图像时计算量小,抗噪性好等优点。通过计算机仿真,表明在在同等条件下,该方法该方法可以有效降低图像噪声、保持图像细节,提高图像质量。
The recognition accuracy would be sharply decreased,when the fuzzy extent of the picture is beyond mild,applying the traditional License Plate automatic identification technology based on Threshold method.Generally speaking,the fuzzy extent is mainly between slight and moderate,and how to solve the problem of the plate recognition in this range is of significance.According to the characteristics of the fuzzy caused by movement,this thesis analyzes the fuzzy image's degradation model and restoration model in uniform rectilinear motion,puts forward PSF model which is suitable for this system,combining with Wiener Filter to restore the images.Because of the advantages of small amount of calculation,strong noise-resistance,etc.,this method is proved to be effective in decreasing the image noise,remaining the image details and improving the image quality under the same situation,through the way of computer simulation.