论文部分内容阅读
为了提高在低比特率条件下的解码图像质量和视觉效果,根据交通图像背景和局部相似特点与原子参数量化特性,给出了基于原子参数预测和量化的交通图像压缩算法。该算法首先分解一批交通图像,以分解后的原子参数构建原子库,在此基础上,利用构建原子库对稀疏分解的原子参数进行预测和量化,然后对投影分量进行排序差分处理,采用变长编码对处理后的投影分量进行编码,根据投影分量的重排顺序,对经过预测和量化后的原子参数进行相应的重排,最后采用算术编码对重排后的原子参数进行编码。仿真试验结果表明,与已有文献中的方法相比,该算法能够更有效的实现交通图像的压缩,在相同压缩比下,解码图像有更高的峰值信噪比和主观图像质量。
In order to improve the decoded image quality and visual effect under low bit-rate conditions, a traffic image compression algorithm based on the prediction and quantification of atomic parameters is given according to the background features of the traffic images and local similarities and the quantization characteristics of atomic parameters. Firstly, the algorithm decomposes a batch of traffic images and constructs an atomic library based on the decomposed atomic parameters. On the basis of this, the atom parameters of sparse decomposition are predicted and quantified by constructing an atomic library. Then, the projection components are sorted and differentiated, The long coding encodes the processed projection components, rearranges the predicted and quantized atomic parameters according to the rearrangement order of the projection components, and finally uses the arithmetic coding to encode the rearranged atomic parameters. Simulation results show that compared with the existing methods, the proposed algorithm can achieve more efficient traffic image compression, and the peak compression ratio and subjective image quality of the decoded image are higher under the same compression ratio.