A harmonic means pooling strategy for structural similarity index measurement in image quality assessment

Structural similarity index measurement (SSIM) is one of the most well known method in full reference image quality assessment metrics (FR-IQA). In this paper, a novel pooling strategy based on harmonic mean is proposed to enhance the performance of SSIM. Instead of arithmetic mean, the proposed pooling by harmonic mean tends to emphasize the contributions from the local severely distorted regions or pixels in the definition of assessment function using reciprocal transformation. The proposed object function has higher correlation with human perception, which is mostly affected with the regions having severely distorted points or regions. In addition, salience information is introduced to the object function for a better consideration of subject visual attention. The proposed pooling strategy is applied to classical SSIM and its variants, GSSIM and FSIM. The experimental results have demonstrated that the FR-IQA metrics with proposed pooling strategy have better performances compared to the standard versions, especially on the images with small but seriously distorted regions.