基于小波分析的图像稀疏保真度评价

该文针对传统的图像质量评价方法无法有效模拟人类视觉系统(HVS)存在的不足，提出基于小波分析的加权稀疏保真度(Weighting Sparse Fidelity, WSF)图像评价算法。算法以模拟人类视觉系统的神经网络为切入点，对图像进行一阶小波分解得到4个不同方向的子带图像，然后将子带图像分成88大小的图像块，采用快速独立分量分析(FastICA)的方法对各个图像块进行训练并提取图像特征检测矩阵，根据特征检测矩阵计算各子带图像块的稀疏特征值并建立稀疏保真度质量评价模型。在此基础上，根据细节信息的不同对低频子带图像进行区间划分并设置视觉权重，使之更加接近人眼的主观视觉。实验中对LIVE库中所有图像进行算法验证，其结果表明，所提方法能很好地对各种失真类型的图像进行评价。基于小波分析的稀疏保真度评价算法能够有效模拟人类视觉系统的多频特性和视觉皮层感知机制，弥补现有图像质量评价方法在此方面的不足。

Sparse Image Fidelity Evaluation Based on Wavelet Analysis

To overcome the limitations of traditional image quality assessment methods, which are not well consistent with subjective human evaluation, a quality assessment algorithm of Weighting Sparse Fidelity (WSF) based on wavelet analysis is proposed. The arithmetic simulates nerve network of Human Vision System (HVS) as research point, the image is decomposed with wavelet into four-sub band images, which are divided into blocks at size of , then using Fast Independent Component Analysis training (FastICA) method to train the image blocks. Then, each image block sparse character matrix is extracted to calculate the sparse feature fidelity of the image and build the sparse fidelity quality evaluation model. On this basis, the image is divided into a plurality of interval according to the different details of the visual image information and a visual weight is set in each section, which can be consistent with subjective human evaluation. The experiment results on LIVE database show that the proposed method has a good evaluation of all kinds of distortion types and is highly consistent with human subjective evaluations. The proposed algorithm can effectively simulate the weighted visual cortex of the human visual system perception mechanisms, which compensates for deficiencies of existing image quality assessment methods.