最小能量约束的图像交互估计去模糊算法

Image deblurring method by alternate estimation with minimum energy constraint

为解决图像模糊引起的车牌识别性能下降问题,提出一种自适应的图像去模糊方法。基于最小能量准则交互估计核函数和最优图像,构建图像质量评价因子判决条件作为迭代终止条件,实现自适应的图像去模糊处理。通过融合锐化指数和自然图像质量评价指标构造图像质量评价因子,用于区分清晰图像和模糊图像,仅对模糊图像进行去模糊处理,并用于控制交互估计的迭代进程。实验结果表明,所提方法能有效区分模糊图像和清晰图像,去模糊效果好,可以提高车牌识别率。

For solving the problem of plate recognition performance degradation caused by image blurring,an adaptive image deblurring method was proposed.Both kernel function and optimal image based on minimum energy principle were alternately estimated,and a decision criterion about image quality evaluation factor was built as the termination condition of iteration,to realize image deblurring.The image quality evaluation factor was constructed by fusing sharpening index and natural image quality evaluation,which was used to distinguish clear and blurred images for executing deblurring operator only for blurred images.It was used to control the iterative process of alternate estimation.Experimental results show that the proposed method can effectively distinguish between blurred images and clear images,and it has good deblurring performance and can improve recognition rate for plate recognition.