A new method for analysing eye-tracking measurement data

Abstract

When it comes to the function of photography in graphic communication, we should be aware of many different aspects of the medium. The standard criterion for a correct and successful photograph should always be its technical and artistic aspect. If something is not right, for example, if the exposure or composition are not correct, a photo should not be used for professional purposes. The third aspect is the content or meaning, where final decisions are usually made. With this in mind, editors are usually presented with photos from which they must select those to be used in a final publication. Since most editorial decisions are made on the basis of content or meaning, which is usually described as communication value, some decisions also involve technical aspects, such as whether it is better to use a slightly underexposed or slightly out-of-focus photo. In these cases, the standard criteria changes because the shooting conditions did not allow the photographer to take a technically correct photo, but the content or meaning is too important not to be published. The research focuses on measuring how the way people see different photos changes when they are not technically perfect. Using eye-tracking technology, where we can measure where a person is looking and for how long, we can get an accurate idea of what that person is seeing and the way their eyes move. This type of measurement is actually not a problem and has been successfully used in many research studies. The main question to be answered in this study was how the nature of image perception changes when the image is distorted in some way. Therefore, a new method for analysing eye-tracking data was developed. The results show that eye-tracking can be used to determine how technical aspects of photography affect the way we look at it. The final judgement that the method works was made by comparing the data with data gathered with subjective tests in which observers had to choose between different distorted images and decide which is more acceptable. The correlation between the results of new method and a subjective testing is very strong.