Subjective and Objective Quality Assessment of Stitched Images for Virtual Reality.

We consider the problem of quality assessment (QA) of image stitching algorithms used to generate panoramic images for virtual reality applications. Our contributions are two-fold. We design the Indian Institute of Science Stitched Image QA (ISIQA) database consisting of 264 stitched images and 6600 human quality ratings. The database consists of a variety of artifacts due to stitching such as blur, ghosting, photometric, and geometric distortions. We then devise an objective QA model called the stitched image quality evaluator (SIQE) using the statistics of steerable pyramid decompositions. In particular, we propose a Gaussian mixture model to capture the bivariate statistics of neighboring coefficients of steerable pyramid decompositions and show this to be effective in modeling the increased spatial correlation due to ghosting artifacts. We show through extensive experiments that our quality model correlates very well with subjective scores in the ISIQA database. The ISIQA database as well as the software release of SIQE has been made available online for public use and evaluation purposes.