AUTOMATIC PRODUCT COLOR DESIGN USING GENETIC SEARCHING

Due to the remarkable advances in numerically–controlled machining technology for product manufacture in recent decades, the functional aspects of many of the consumptive products used in our daily lives are now fully matured. Subsequently, for enterprises seeking to develop new products in today’s highly competitive marketplace, typically characterized by short product life cycles, the apparent style of a product, i.e. its form and color, is of an ever-increasing importance. Designing and manufacturing the wide variety of product forms required to meet the diverse requirements of individual consumers is both time consuming and expensive. However, by varying the color combinations of the visible components of a product, an enterprise can generate a wide variety of different product image perceptions. In this way, a product produced from a single fixed mold can be offered to the market in numerous color combinations so as to satisfy the individual consumers’ various needs.

Color is an important factor for the appearance of a product. The designed colors can be digitally simulated on a developed product by using CAD systems. Although many developed color-harmony theories have been applied in the color planning stage of product design, most of them were developed based on the object color system with the aesthetic view and cannot be appropriately quantified in digitally light-based form. However, the representations produced in computer-based color simulations are generated mainly using the CIE color system. Accordingly, this study utilizes Moon and Spencers’ color harmony theory to develop light-based quantitative aesthetic measurement. Besides, a linguistic evaluation method using gray theory is also performed on the designed colors. In an inverse process, genetic algorithms are applied to search for a near-optimal color combination which satisfies the designer’s required product-color linguistic image on a qualified aesthetic level. The proposed system comprises two sub-systems, namely an image prediction sub-system and a color-combination search sub-system. In the proposed approach, color parameters are input to the image prediction sub-system and are output from the color-combination search sub-system. Meanwhile, the predicted/desired image evaluation is output from the image prediction sub-system and is input to the color-combination search sub-system. This automatic design system enables designers to rapidly simulate the designed colors of a product and then obtain its corresponding image evaluation, or to search for the ideal color combination which generates the required image perception.