Application of back-propagation neural network-based approach to icon image design

A satisfactory iconic interface or icon can evoke users' image perception, increase their interest, and promote them to interact with the specific applications. It is necessary to develop an approach to examine the correlation between the icon design and users' image perceptions. This study proposed a back-propagation neural network (BPNN)-based approach to icon image design. A series of icon image evaluations are conducted to examine the relationship between the characteristics of static icon and its associated users' image perception. The BPNN model is constructed to transform the visual effect of static icon effects into a corresponding set of users' image perceptions for providing designers with the ability to predict the likely user image perception when presented with a particular icon and to make an appropriate design decision.