Image statistics of melanopsin-mediated signals

Poster Presentation 33.337: Sunday, May 19, 2024, 8:30 am – 12:30 pm, Banyan Breezeway
Session: Color, Light and Materials: Neural mechanisms, models, disorders

Pablo Barrionuevo1,2 (), Francisco Diaz Barrancas1; 1Justus Liebig University, 2ILAV, CONICET-UNT

Melanopsin-expressing intrinsically photosensitive retinal ganglion cells (ipRGCs) project to nervous areas involved in visual processing. It is thought that stable and long-term responses of these cells might provide unique information to visual perception complementary to the transient and adaptive rod and cone responses. There is a close match between natural statistics and processing of contrasts in the early visual system but most of the studies were carried out considering luminance (cone) statistics. This study aimed to determine the intensity and contrast statistics of melanopsin and ipRGC codification of photoresponses in natural and human-made scenes. We have computed intensity (melanopsin excitation and luminance) and within-image Michelson contrasts using the melanopsin (Mel) and cone (L, M, and S) CIE fundamentals on hyperspectral images with natural (21) or human-made elements (10) under natural illumination. Statistics were obtained considering 16 patches per image, mimicking receptive fields, each one weighted by a raised cosine window covering 2.4° (ipRGCs) and 1° (luminance-processing parasol cells). Also, statistics on inferred ipRGC photoreceptor codification (Mel+L+M-S) were computed. Our findings, based on median values, reveal that human-made environments exhibit significantly higher melanopsin (2.53x) and ipRGC (15.3x) excitations compared to natural environments. This increment is higher than the luminance increment (1.57x). Regarding median contrasts, lower values for melanopsin (41.2% and 41.3%) and ipRGCs (42.3% and 42.6%) than luminance (49.1% and 45.9%) were found for both natural and human-made environments, respectively. These differences spread all over the contrast range. There was no apparent correlation between melanopsin contrasts and excitations. Our analyses showed that the melanopsin-mediated and ipRGC-mediated statistics differ from luminance-mediated statistics. Including human-made elements in the scene affects melanopsin excitations more than luminance, but melanopsin contrast was not affected. Melanopsin statistics found in natural environments might serve as a reference to discuss the implications of lab-conducted studies.

Acknowledgements: This study has received funding from the German Research Foundation (DFG) - project number 222641018 - SFB/TRR 135 TPs C2 and B2 and by the European Research Council Advanced Grant ‘An object-oriented approach to color: Color 3.0.’ -project number 884116.