Purchase this article with an account.
S.I. Guthrie, T. Simpson, J. Varikooty, D. Fonn; Background Subtraction and Contrast Enhancement for Interferometric Images of the Human Corneal Tear Film . Invest. Ophthalmol. Vis. Sci. 2006;47(13):2399.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Interference of visible light is a common tool for observing the human tear film, however quantitative analysis of interferometric data is hindered by the low contrast of the recorded fringes (and the background anatomy shown). We attempt to improve on this through frame by frame analysis of video.
We use the freely–available ImageJ programming tool to implement a background subtraction algorithm to interferometric images of the human precorneal tear film. The algorithm removes static components and leaves dynamic components The resulting images are then contrast–enhanced, and compared back to the original data (also contrast–enhanced). The background subtraction algorithm uses a boxcar–type frame–average of varying width to remove dynamic effects. Images are evaluated using the Michelson contrast.
Our routine removes most of the background image from the video frames, leaving the tear film exposed. The image with clearly visible iris is pre–analysis, and the second is with background subtraction – both are contrast–enhanced. Contrast enhancement of the post–processed images was greater than 40% in the images analysed.
Freely–available software and relatively simple routines have been shown to produce dramatic improvements in the contrast of interference fringes in data from human eyes. The boxcar routine presents the best improvement, though the number of frames varies depending on the data. The algorith was designed to analyse the dynamic tear film and once the film is stationary, the algorithm is no longer applicable.
This PDF is available to Subscribers Only