摘要
由于采集到的车牌图像分辨率较小,为获取高分辨率的图像,提出基于马尔可夫随机场模型的车牌图像超分辨率复原算法。首先对图像进行分块;然后利用马尔可夫随机场对这些分块进行建模,通过模型学习训练库中高低分辨率图像的关系,预测待复原的低分辨率车牌图像的高频细节信息。实验结果表明,本算法对车牌图像取得较好的复原效果,算法复原的超分辨率车牌图像更接近于真实图像,具有更高的峰值信噪比。
Since the license plate images obtained have low resolution, in order to obtain high resolution images, the paper suggested MRF based super-resolution algorithm to recovery license plate images. First divided the image into blocks, and then used MRF to model these blocks. Through the relationship between high and low resolution images in learning library, presented the high-frequency details in the low resolution image. The experimental results show that Markov random field model based super-resolution image registration algorithm obtains a better recovery results, the super-resolution license plate image is closer to the real image, with a higher peak signal to noise ratio.
出处
《计算机应用研究》
CSCD
北大核心
2010年第3期1170-1172,1186,共4页
Application Research of Computers
基金
国家教育部重点资助项目(107094)
四川大学青年教师基金资助项目
关键词
图像复原
马尔可夫随机场
基于学习的超分辨率
最大后验概率
image restoration
Markov random field(MRF)
learning based super-resolution
maximum a posterior (MAP)