Contrast is the contradiction in luminance or colour that makes an object distinguishable, and is determined by the difference in the colour and brightness of the object and other objects within the same field of view. The human eye shows high sensitivity towards contrast and it is an important aspect of vision and us being able to see "well".
Image contrasting is the process of changing the contrast spectrum of an existing image to focus it the band which is best suited for human vision, which allows us to clearly distinguish the components of the image.
In the case of low-vision people, image contrasting turns out to be one of the major factors that can impact their ability to see. A well contrasted image will allow them to view the contents of the image well, and if there's text, allow them to read it better.
We employ three broad techniques for image contrasting:
Contrast by Stretching
Histogram Equalization
Regular Equalization
Adaptive Equalization
Novel "Brightness Differential" Technique
Fig: Image of Reindeer Before Contrasting
Fig: Image of Reindeer After Contrasting
We use two empirical metrics and one observational metric for evaluating our image contrasting techniques on images:
Empirical metrics:
Histogram + Cumulative Distribution Function (CDF) of Pixel Intensities: Ideally, we want our histogram height to be uniform, rather than there being highly concentrated regions for some intensity bins.
Performance of OCR on Contrasted Image
Observational metric:
How Effective the Contrasting Technique seems to us as Observers
Fig: Histogram and Cumulative Distribution Function Plot Before and After Contrasting by Stretching