Visual performance-based image enhancement methodology: an investigation of three Retinex algorithms
Kelly E. Neriani, Travis J. Herbranson, Alan R. Pinkus, Christine Task, H. Lee Task · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2005
While vast numbers of image enhancing algorithms have already been developed, the majority of these algorithms have not been assessed in terms of their visual performance-enhancing effects using militarily relevant scenarios. The goal of this research was to develop a visual performance-based assessment methodology and apply it to assess three Retinex algorithms. The image enhancing algorithms used in this study are the two algorithms described in Funt, Ciurea, and McCann as McCann99 Retinex and Frankle-McCann Retinex, and the multiscale Retinex with color restoration (MSRCR) algorithm. This paper discusses the methodology developed to acquire objective human visual performance data as a means of evaluating various image enhancement algorithms. The basic approach is to determine whether or not standard objective performance metrics, such as response time and error rate, are improved when viewing the enhanced images versus the baseline, non-enhanced images. Four observers completed a visual search task using a spatial-forced-choice paradigm. Observers had to search images for a target (a military vehicle) hidden among foliage and then indicate in which quadrant of the screen the target was located. Response time and percent correct were measured for each observer. Future directions and the viability of this technique are also discussed.