A comparative study of methods to explore searchers' affective perceptions of images

Introduction. Previous search query analysis studies demonstrated that emotional perceptions and needs tend not to be sufficiently reflected in image queries. To provide a better understanding of the emotional attributes of an image, this study explored and compared such attributes obtained from related tasks. Method. This study analysed attributes of an image through sorting, describing and searching. A data­set was obtained from Flickr images tagged with one of six basic emotions (love, joy, surprise, anger, sadness and fear); fifty­nine students participated. Results. The sorting and describing tasks revealed two categories of emotions, joy and sadness, which people perceive from images. They also revealed some emotional terms which were chosen frequently as labels for sorted images. Comparison of image perceptions obtained from the three tasks performed on an image demonstrated similar perceptions, but there are minor differences resulting from the unique characteristics of each task or research setting. Conclusions. The categories of joy and sadness, along with the identified frequently used emotional terms, should receive attention in designing image indexing and retrieval systems. Also, the features of each task and environment setting should be considered when developing a research design exploring emotional aspects of an image.