Color Theme--based Aesthetic Enhancement Algorithm to Emulate the Human Perception of Beauty in Photos

Fine Art Photography is one of the most popular art forms, which creates lasting impressions that elicit various human emotional reactions. Photo aesthetic enhancement aims at improving the aesthetic level of the photo to please humans by updating color appearance or modifying the geometry structure of objects within that photo. Even though several aesthetic enhancement methods have been proposed, to our knowledge, there is no research to explore, highlight, and accentuate photos’ intrinsic aesthetic value to elicit a stronger response from the human observer about the photos’ theme. To meet this challenge, a new multimedia technology called automatic color theme--based aesthetic enhancement (CT-AEA) is proposed by leveraging big online data to perform timely collection and learning of humans’ current aesthetic perception-behavior over photos and color themes in art, fashion, and design. Unlike existing aesthetic enhancement that examines the composition, such as the geometric structure of the image contents and color/luminance-related (color tone and luminance distribution) characteristics, this CT-AEA takes into consideration the importance of a suitable color theme, namely a set of dominant colors for the design when assessing the aesthetic appearance of a photo. This algorithm is composed of (1) utilizing the knowledge gained from the human evaluator's perception of beauty from existing online datasets, rather than simply applying prior existing knowledge of color harmony theory; (2) developing a new color theme difference equation that exhibits order-invariance and percentage-sensitive properties; (3) designing an optimal color theme recommendation to maximize the aesthetic performance, while minimizing the color modification cost to solve the problems of color inconsistencies and distortion. Experimental results, quantitative measure, and comparison tests demonstrate the algorithm's effectiveness, advantages, and potential for use in many color-related art and design applications.