欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
根据小波变换的高频分量包含边缘及纹理等人眼比较关注的细节信息,提出了一种基于小波变换的高频重构的立体图像质量评价方法,包括左右视点质量评价和立体感评价两部分。通过模拟人类视觉系统处理原始和失真的左右视点图像,利用结构相似度算法得到左右视点质量评价值;用加权处理后的小波高频分量经过重构得到人眼关注的细节信息图,分别计算原始和失真的左右视点细节信息图的绝对差值图,利用结构相似度算法对两幅绝对差值图评价得到立体感质量评价值。将两种质量评价值进行加权融合得到最终的立体图像质量评价值。整体实验结果的Pearson线性相关系数在0.94以上,Spearman秩相关系数值在0.93以上,均方根误差值接近5.5,能较好地符合人眼视觉特性。
英文摘要：
According to the high frequency sub-bands of wavelet decompression of signals contain much important detail information such as edge and texture features, a new stereo image quality assessment algorithm is proposed based on wavelet decompression of signals. The new algorithm combines the left-right image quality assessment and the stereo- scopic perception quality assessment. The left and right views~ quality is evaluated by making use of structural similarity image metric （SSIM） after model human visual system. The high frequency sub-bands of wavelet coefficients are weigh- ted to restructure new images to compute the original and the distortion stereo detail map＇s disparity maps. Results of left-right image quality and the stereo perception quality are combined to describe the final quality of stereo images. By applying the proposed model to stereo image test database, experimental results demonstrate that the overall Pearson linear correlation （PLCC） indicator of the proposed algorithm reaches 0. 94, and it Spearman rank order correlation coefficient （SROCC） indicator reaches 0. 93, and its RMSE approaches to 5.5, which indicate that the proposed algorithm is consisted with human perception well.
同期刊论文项目
同项目期刊论文
期刊信息
位置：