A Depth Dependent Quality Metric for Evaluation of Coded Integral Imaging Based 3D-Images

The two-dimensional quality metric peak-signal-to-noise-ratio (PSNR) is often used to evaluate the quality of coding schemes for integral imaging (II) based 3D-images. The PSNR may be applied to the full II resulting in single accumulate quality metric covering all possible views. Alternatively, it may be applied to each view results in a metric depending on viewing angle. However, both of these approaches fail to capture a coding scheme's distribution of artifacts at different depths within the 3D-image. In this paper we propose a metric that determines the 3D-image quality at different depths. First we introduce this ID measure, and the operations that it is based on, followed by the experimental setup used to evaluate it. Finally, the metric is evaluated on a set of 3D-images; each coded using four different coding schemes and compared with visual inspection of the introduced coding distortion. The results indicate a good correlation with the coding artifacts and their distribution over different depths.