Evaluation of changes in image appearance with changes in displayed image size
Image appearance is affected by various factors, including the output imaging system, the viewing conditions, the physical changes in image size which correspond to changes in the angle subtending the observer’s eye. Many studies have been conducted to identify and quantify the changes in image appearance with respect to the image size/viewing angle. Most of them have been using either colour patches, or artificially generated patterns.
In this research project, a series of paired-comparison experiments have been conducted to identify image attributes which are most affected by changing the physical displayed image size, on an LCD display. Sharpness, contrast, brightness, colourfulness, and noise were found to be affected, but to different degrees. But it was found that there was no effect on hue. Similar results were found in a recent study by Wang et al, who investigated the changes in appearance of all image attributes as well as image compression with changes in the visual angle. The two most affected image attributes were first sharpness and then contrast.
For the evaluation and modeling of sharpness changes with varying image size, a set of images with varying sharpness level was required. The ISO’s softcopy ruler method was adopted to create a series of frequency domain sharpness filters by taking into account the combined imaging system MTF and the viewing (or observation) distance. Filters were created to have a constant interval in just-noticeable differences (JNDs) to obtain perceptually calibrated results. A total of 72 filters spaced with an interval of same JNDs in sharpness were created.The results obtained from the experiments showed that the effect of changing image size on perceived image sharpness was scene dependent. Therefore, scene analysis techniques using statistical measures have been carried out to classify scenes into several categories in order to calibrate the ruler images to the scenes. A similar methodology is pursued for the evaluation of changes in perceived contrast will follow in the remainder of research.
The relationship between objective display measures and visual appearance I also sought in an effort to expand iCAM (image colour appearance model) so that it accounts for varying image size. Relevant modules from the iCAM will be used and possibly extended, while additional modules will be built for the purpose.