Abstract: At low bit rates，the Block Discrete Cosine Transform（BDCT） compression methods often give rise to blocking artifacts.In various approaches to the problem，Quantization Constraint Set（QCS） is widely used as a priori information about the coding process.In this paper，we propose a modified constraint set based on the statistical property of quantization errors.The experimental results show that while combined with typical POCS-based algorithms，the proposed QCS outperforms traditional QCS and produces results comparable with those achieved by narrow quantization constraint set（NQCS）.

Key words: blocking artifacts, Block Discrete Cosine Transform（BDCT）, Projection Onto Convex Sets（POCS）, Quantization Constraint Set（GCS）

摘要： 低比特率时，基于块离散余弦变换（BDCT）的图像压缩方法常常会引起块效应。量化约束集（QCS）作为关于编码模型的重要先验信息在块效应消除方法中被广泛地采用。利用量化误差的统计特性，提出了一种改进的量化约束集（MQCS）。实验结果表明当在现有的基于凸集投影（POCS）的去块算法中以新的约束集代替传统量化约束集，峰值信噪比（PSNR）和主观图像质量有不同程度的提高，性能与窄量化约束集（NQCS）相当。

关键词: 块效应, 块离散余弦变换, 凸集投影, 量化约束集