HDR VDP is software that can predict perceptual differences between two images. It is a useful tool for evaluating the quality of high dynamic range (HDR) images.
Visual difference metrics are essential in evaluating whether differences between two images are visible to the human observer or not. These metrics find application in testing the visibility of important visual information or the visibility of noise.
Main Section:
One such metric is the Visual Difference Predictor (VDP), which processes two input images (a reference image and a distorted image) to produce a probability of detection map. This map indicates the likelihood of us noticing a difference between two images for each part of an image. The probability of detection map uses red and green colors to denote high and low probability, respectively.
Furthermore, the probability of detection map indicates potential distortions in the image. For instance, the red color predominates in snow-covered paths where there is a smooth texture. As such, visual masking is minimal, and distortions are more prominent.
While there are numerous visible difference metrics that serve a similar purpose, the High Dynamic Range Visible Difference Predictor (HDR-VDP) has two unique advantages over other metrics. Firstly, it works with the full range of luminance values that can be found in real-world scenarios, including high dynamic range (HDR) images. Secondly, we provide a complete source code for free.
The HDR-VDP can work within the entire range of luminance values that the human eye can perceive. It serves as an input to the metric, with the algorithm taking into account the aspects of high-contrast vision like nonlinear response to light and local adaptation, among others.
Conclusion:
In summary, the Visual Difference Predictor, and specifically the High Dynamic Range Visual Difference Predictor, is an essential tool in evaluating the visibility of noise and important visual information in an image. Moreover, the HDR-VDP is unique in its ability to work with the full range of luminance values and its provision of complete source code for free.
Version 1.6: N/A