%0 Journal Article
%T 工业机器人视觉图像的质量评价方法研究<br>Research on Image Quality Assessment for Industrial Robot Vision System
%A 王靖宇
%A 王霰禹
%A 张科
%A 张彦华
%A 王景鹏
%J 机械科学与技术
%D 2018
%X 工业机器人视觉系统在校准和标定过程中需要对成像质量进行客观评价，而在参考图像已知的情况下，现有图像质量评价方法无法有效表征图像局部结构差异，难以充分考虑像素级差异对局部结构的影响。针对上述问题，提出了一种基于视觉显著性的全参考图像质量评价方法。其中，设计了一种基于梯度图的差异度表征方法，利用相位一致性和梯度图来描述图像局部结构和像素级的差异，进而提出了基于视觉显著性的图像质量评价方法，通过利用像素级差异对图像局部结构进行权重赋值，从而提高算法对图像进行质量评价时的性能。实验结果表明，本文方法和其他四种典型图像质量评价算法相比，能够在三个标准图像数据库上取得更高的质量评价分数。<br>Industrial robot vision system need to objectively evaluate image quality in calibration process. While the reference image is known, the existing image quality assessment (IQA) method cannot effectively characterize the local structure difference, thus it is difficult to fully consider the impact of pixel level difference. Therefore, the full reference IQA method based on the visual saliency is proposed in this paper. To be specific, the difference characterization method is designed, in which the phase congruency and gradient map are used to describe the differences in local structure. Thus, the IQA method based on the visual saliency is proposed, in which the pixel level difference weights are assigned to the different local structure of image in order to improve the performance. The results show that the present method can achieve the quality evaluation scores in three standard databases
%K 工业机器人
%K 机器视觉
%K 图像质量评价
%K 全参考
%K 视觉显著性&lt
%K br&gt
%K double tree complex wavelet
%K morphological
%K telemetry data
%K data filtering
%K abnormal data
%U http://journals.nwpu.edu.cn/jxkxyjs/CN/abstract/abstract6992.shtml