摘要： 目前有些图像质量评估方法未考虑到感兴趣区域对图像质量的影响，所得结果与人的主观感觉有误差。为此，提出基于感兴趣区域的双尺度图像质量评价方法。将图像质量的评价分为感兴趣区域和不感兴趣区域的加权组合评价，在感兴趣区域内使用双尺度图像质量评价方法，不感兴趣区域使用传统的结构相似度评价方法。实验结果表明，该方法考虑了感兴趣区域对图像质量的影响，强化了感兴趣区域评价值在总评价值中的比重，更符合人的主观视觉质量。
关键词:
医学图像,
结构相似度,
质量评估,
显著性,
非线性回归拟合
Abstract: Currently, some image quality assessment methods do not consider the impact of the Region of Interest(ROI) of the image quality, there is deviation with human subjective visual quality. In this paper, a strategy is proposed based on ROI and dual-scale edge structure similarity. The quality assessment of the image is a weighted combination of ROI and NON-ROI, the dual-scale edge structure similarity is used in ROI, and the classical structure similarity is applied in NON-ROI. Experimental results show that, considering the influence of ROI to the image, strengthening the evaluation value of the ROI in the proportion of the overall rating value, the model is more consistent with human’s subjective visual quality.
Key words:
medical image,
structure similarity,
quality assessment,
significance,
nonlinear regression fitting
中图分类号: