Skip Navigation Links
ServicesExpand Services
ProductsExpand Products
ResearchExpand Research
MembershipExpand Membership
About Us
Contact Us




We conduct human vision research  to understand visual performance,  to develop computational models to simulate biological visual image processing, and to establish image quality evaluation methods based on human visual perception.

Specific research projects are launched based on our interests and the needs for potential industry applications.

Currently, we are developing a general-purpose computational algorithm to simulate visual detection performance in order to  estimate the target visibility and the perceived brightness of visual patterns seeing by an average observer in diverse viewing conditions.  In the model development, we included well-known visual properties, also a concept of  implicit masking.  You can find the basic framework for this model at this link.  For more related works, please check Jian Yang's publication list.


(1) August, 2006:  A paper entitled "Simulating visual pattern detection and brightness perception based on implicit masking" has been accepted  for a publication in EURASIP Journal on Applied Signal Processing. (PDF file)
(2) October 2006: A poster, entitled "Modeling Modelfest Data and Luminance Dependent CSFs Based on Implicit Masking" was presented at OSA Fall Vision Meeting, Rochester, NY. (PDF file)


(1) Simulating Human Visual Performance with Eye Diseases (looking for sponsors or collaborators)
Abstract: The proposed research is to develop computational tools to model and simulate human visual performance in order to evaluate the efficiency of different psychophysical tests in screening eye diseases. Deterioration in visual functions can signify eye diseases, such as glaucoma, retinopathy, age-related macular degeneration, cataract, etc. Psychophysical tests, such as visual field tests, have been widely used in clinics for screening and monitoring glaucoma patients. In this area, investigators are striving to design better tests that could pick up the weak signs of early stage eye disease development. Usually, the effectiveness of a new test requires experimental tests with a large sample of patients and normal observers. Such an approach is costly and time consuming. The tools to be developed in the proposed research are to reduce the need to use human observers, by using a computational model observer and simulated disease models. The model will be used to optimize new tests to be evaluated by use in the clinical trials, thereby saving the time now spent evaluating suboptimal tests. In case that a low cost, efficient, and easy to use perimetry test is available, we can expect that a large population screening service for glaucoma will be feasible and cost effective. Phase I of the project aims at developing a normal achromatic foveal model observer that is general enough to work under a variety of test conditions. This model observer will include functional components that can vary with eye diseases. If phase I is successful, the phase II goal is to extend the model observer to estimate visual detection performance in peripheral vision, temporal and chromatic visual image processing, and to simulate one eye disease model: glaucomatous ganglion cell loss. A method will be developed to capture the variability of visual performance in order to assess the sensitivity of a test with a fixed number of trials, and experiments will conducted to validate the model estimated visual performance with real patients.