有效保留图像细节的模糊区域复原方法仿真

Fuzzy Area Restoration Method Simulation for Preserving Image Details Effectively

平滑区域阶跃效应是影响模糊图像复原的重要影响因素之一。为了有效提高图像模糊区域复原的质量,针对当前方法复原后存在平滑区域阶跃效应而无法有效保留图像细节信息问题,提出基于广义逆矩阵的有效保留图像细节的模糊区域复原方法。对测试模糊图像进行分块处理,利用SIFT算法在图像模糊区域与清晰区域之间进行块配准,计算获得各个图像块的质心坐标。采用质心法计算每个图像块发生偏移的角度及距离,综合所有图像块的偏移结果,估计出图像模糊区域的点扩散函数。以点扩散函数的矩阵形式构建图像模糊区域复原模型,运用零边界像素条件的保护边缘特性克服平滑区域阶跃效应的优势,完善图像模糊区域复原模型。实验结果表明,所提方法对模糊区域进行复原能够有效保留图像细节信息,且相比现行复原方法峰值信噪比增量提高了2. 4~3. 4dB。

In order to effectively improve the restoration quality of image in fuzzy region, a method of fuzzy region restoration to effectively preserve image details based on a generalized inverse matrix was proposed. Firstly, the tested fuzzy image was divided into blocks. Secondly, SIFF algorithm was used to perform block registration between fuzzy area and clear area of image. Then, the center-of-mass coordinate of each image block was calculated. Moreover, the centroid method was used to calculate the angle and distance of offset of each image block. Based on offset results of all image blocks, the point spread function of image fuzzy region was estimated. Meanwhile, the matrix form of point spread function was used to build restoration model of image fuzzy region. Finally, the advantage that the feature of preserving edge of zero boundary pixel condition could overcome the step effect in smooth region was used to improve and perfect the restoration model of image fuzzy region. Simulation results show that the proposed method can effectively preserve details of image through recovering the fuzzy region. Compared with current restoration methods, the peak signal-to-noise ratio is increased by 2.4 - 3.4 dB.