基于优先权和匹配度量的图像修复算法

Image Inpainting Algorithm Based on Priority and Matching Measurement

目的针对当前图像修复算法主要通过固定单一模板大小实现修复块与匹配块之间的匹配度量来完成图像复原,导致其存在一定的模糊效应以及振铃效应等不足,这里提出基于改进优先权和匹配优化度量的图像修复算法。方法首先,利用数据项构造平滑因子,建立优先权模型,度量待修复像素点的优先权,选定优先修复块。然后,制定四级模板大小,利用误差平方和函数,结合模板大小特性,构造匹配度量模型,用于实现修复块和匹配块之间的动态匹配,选取最优匹配模板,对待修复块进行填充修复。最后,利用待修复像素点及其邻域像素点的灰度值构造邻域灰度差分模型,用于对修复区域的边缘进行缝合,优化修复效果。利用最优匹配度量结果,构造置信度更新模型,对置信度项进行更新,实现图像修复。结果仿真实验结果显示,与当前图像修复算法相比,所提算法具有更高的修复质量,其输出图像无模糊效应与振铃效应。结论所提算法能够较好地对损坏图像进行复原,在图像信息处理领域具有一定的参考价值。

The work aims to propose an image inpainting algorithm based on priority improvement and matching optimization measurement,with respect to the problem that the current image inpainting algorithm achieves the matching measurement between repair block and matching block to complete the image restoration mainly by the fixed single template size,thus causing such deficiencies as certain blurring effect and ringing effect.First,the smoothing factor was constructed by the data item,the priority model was established to measure the priority of the pixel to be repaired,and the priority repair block was selected.Then,the four level template size was formulated.By means of square error and functions,combined with the template size characteristics,the matching measurement model was constructed to achieve the dynamic matching of repair block and matching block.The best matching template was selected to fill and restore the block to be repaired.Finally,the neighborhood gray differential model was constructed by the gray of the pixels to be restored and their neighborhood pixels,which could be used to stitch the edges of the repaired region and optimize the restoration effect.With the best matching measure,a confidence updating model was constructed to update the confidence item and realize the image restoration.The simulation results showed that,compared with the current image inpainting algorithm,the proposed algorithm had higher inpainting quality,and its output image was subject to no blurring effect and ringing effect.The proposed algorithm can better restore the damaged image and it has certain reference value in the field of image information processing.