Data for PseudoSSIM model for sub-block (4x4, 8x8) prediction using covariance in H.264/AVC

Data for PseudoSSIM model for sub-block (4x4, 8x8) prediction using covariance in H.264/AVC Code book: "BlockSize","SSIM","OneMinSSIM1k","Covariance","AbsMeanDiff","PseudoSSIM","TradDist","OneMinSSIM","LogCovar" "BlockSize", 8 for 8x8, 4 for 4x4 "SSIM", Structural Similarity (SSIM) measure "OneMinSSIM1k", 1-SSIM times 1000, to make it an integer "Covariance", statistical component of SSIM "AbsMeanDiff", additional measure of calculating difference between original and reconstructed "PseudoSSIM", proposed measure "TradDist", L1 norm of Sum of Absolute Transform Difference, (SATD) "OneMinSSIM", 1-SSIM used make SSIM score zero based "LogCovar", Log value of covariance scaled against a maximum covariance score observed, ~13k.本数据集用于基于协方差的H.264/AVC码本子块（4x4、8x8）预测的伪SSIM（PseudoSSIM）模型，包含以下数据字段：1. 块大小（BlockSize）：取值8对应8x8子块，取值4对应4x4子块；2. 结构相似性（Structural Similarity, SSIM）指标：经典图像质量评估指标SSIM的得分；3. 1000倍1减SSIM（OneMinSSIM1k）：将1减去SSIM的结果乘以1000以转换为整数形式的指标；4. 协方差（Covariance）：SSIM的统计组成分量；5. 平均绝对差（AbsMeanDiff）：用于计算原始图像与重构图像间差异的额外评估指标；6. 伪SSIM（PseudoSSIM）：本次研究提出的新型质量评估指标；7. 传统距离（TradDist）：绝对变换差之和（Sum of Absolute Transform Difference, SATD）的L1范数；8. 零基准SSIM（OneMinSSIM）：即1减去SSIM的结果，用于将SSIM得分归一为以零为基准的形式；9. 对数协方差（LogCovar）：以观测到的最大协方差得分（约13000）为缩放基准的协方差对数值。