Computational aesthetics explores how to use computational techniques to predict humans' emotional responses to visual stimuli, thereby enabling computers to imitate humans' aesthetic processes and automatically generate designs that meet the image aesthetic qualities. Aesthetic subjectivity defines the evaluation of image aesthetic qualities as a very challenging task. This study combines aesthetic principles in design with computable image features and, via optimizing a series of key problems, integrates experts' prior knowledge in visual presentation, text semantics, design principles, and cognitive understanding into the multimedia computing framework that facilitates computable automatic design.