Automatic design support and image evaluation of two-coloured products using colour association and colour harmony scales and genetic algorithm

Abstract

Colour plays a key role in determining a consumer’s response to a product’s appearance. Accordingly, this study proposes an automatic design support system which enables a designer either to emulate the colour scheme of a two-coloured product and then to determine its corresponding image perception, or to search for the two-colour combination which most closely meets the required image perception. The proposed system combines a gray theory-based colour-association evaluation method and a colour-harmony-based aesthetic evaluation method to design and evaluate different product-colour schemes. Since colour-harmony theories cannot be implemented directly using the additive primaries (i.e. R (Red), G (Green) and B (Blue)) used in most computer-based colour emulations, this study develops a RGB-based colour-association and colour-harmony measurement scheme to evaluate the image perception of a particular product-colour scheme. In an inverse process, a genetic algorithm is applied to search for the near-optimal colour combination which satisfies the specified product colour-association goal and achieves a high degree of colour harmony. Various case studies involving the design and evaluation of a two-coloured thermos flask are provided for illustration purposes to demonstrate the effectiveness of the proposed method.

Introduction

Nowadays, the apparent style of a product, i.e. its form and its colour, plays a crucial role in determining the likelihood of its success in today’s increasingly competitive marketplace. Designing and manufacturing the wide variety of product forms which would be required to satisfy the diverse requirements of every individual consumer is virtually impossible. However, by varying the colour combinations of the external components of a product, an enterprise can generate a wide variety of different product image perceptions. Adopting this approach, a product produced from a single mould can be offered in numerous colour combinations so as to satisfy the diverse preferences of the consumer group [1].

Colour plays an important role in determining the appeal of a product to its potential customers [2]. Products offered with a single or standardized range of colour combinations are unlikely to satisfy the diverse needs of individual consumers. Therefore, developing products with particular colour schemes targeted at specific consumer groups has emerged as an essential strategy for many enterprises. To support designers in implementing this strategy, a requirement exists for design support systems capable of accurately modelling the relationship between the colour scheme of a product and the likely consumer response.

In his studies of product-colour image perception, Hsiao [3] applied fuzzy set theory to select the most suitable colour for a car from a specified colour range. In a later study [4], the same author once again used fuzzy set theory to establish the correlation between specified colours and corresponding image words. Taking a mobile phone for illustration purposes, Lai et al. [5] used Quantitative Theory Type I and neural networks to examine the individual and combined effects of a product’s colour and form on its overall image perception. Their results showed that product colour has a greater effect on consumer perception of a product than the product form. Marshall et al. [6] surveyed the role of packaging colour in product selection among preschoolers and reported a high correlation between the individual preschooler’s favorite colour and choice of product. Barkat et al. [7] indicated that the use of certain colour additives in cosmetics induced strong feelings of relaxation, excitation, perceived pleasantness and emotional arousal in potential consumers. Choo and Kim [8] investigated the effects of colour in determining the appeal of fashion fabric products using Munsell and PCCS colour notations for the colour variables. Ishihara et al. [9] employed self-organizing neural networks to establish the correlation between colour and describing adjective words in developing an automatic builder for a Kansei Engineering [10] expert system. However, despite the notable contributions of the studies outlined above, the results related only to the problem of single-colour image evaluation. In practice, however, consumer products are generally characterized by colour schemes comprising multiple colours. Hence, the practical application of the studies above is rather limited.

Ou et al. [11] investigated the colour preference aspect of two-colour combinations, and focused particularly on the relationship between colour preference and colour harmony. Meanwhile, the current author combined gray system theory and neural networks to evaluate the multicolour image, the form image and the overall (i.e. colour + form) image of a product [12]. The results of a case study involving the colour design of a door lock and handle showed that the overall product image perception was dominated by the product’s colour rather than by its form. However, neither study considered the effect of colour-harmony in determining a consumer’s response to a product. Lyons and Moretti [13] presented a mathematical colour harmony model for computer interfaces based on Munsell’s aesthetic rules. Hård and Sivik [14] used the Colour Gestalt method to construct a descriptive model for colour combinations based on the NCS colour system. Chuang and Ou [15] investigated the influence of the interval between two colours on the overall colour harmony effect and compared their results with those obtained using Moon and Spencers’ theory. Finally, Shen et al. [16] proposed a linguistic-based evaluation model specified in terms of the CIE colour system for evaluating the harmony characteristics of images comprising multiple colours in the interior design field.

The various colour-image studies outlined above focus primarily on the image evaluation of products with predefined colour combinations. However, comparatively little attention has been paid to the inverse case, i.e. the identification of colour combinations which satisfy a specified colour image evaluation requirement. In general, colour-association orientated schemes designed to facilitate the automatic identification of optimal colour combinations must be supervised by appropriate colour harmony theories. The problem of developing quantitative measures for the evaluation of aesthetics was originally considered by Birkhoff [17]. Subsequently, Moon and Spencer [18], [19], [20] applied Birkhoff’s theories of aesthetic measurement to the problem of colour harmony, where the individual colors were defined using Munsell system variables.

The present study combines a colour-harmony-based aesthetic measurement method with the gray-theory-based colour-association evaluation method proposed by the current author in [1] to evaluate the image sensations induced by a product rendered using different two-colour combinations. In general, the RGB colour system provides a convenient means of rendering different colour schemes on a 3-D CAD model during the colour planning stage. However, the colour-image evaluation functions provided by typical design applications are generally based on colour-appearance systems [21] such as the Munsell system or PCCS (Practical Colour Coordinate system) [3], [4], [8], [22], rather than on additive colour systems. However, in the design support system developed in the current study, the Munsell-based colour parameters of the colour-harmony measurement scheme are transformed into additive primaries (i.e. red, green and blue) such that the evaluation results can be integrated with a CAD system. Furthermore, in an inverse process, a genetic algorithm [23] is employed to search for the colour combination which most closely satisfies the required colour-association and colour-harmony requirements for the product.

Unlike conventional problem-solving techniques, genetic algorithms converge toward the optimal solution from multiple directions. In the current application, the chromosomes produced during the population-based search procedure represent the colour-design candidates produced by designers during a conventional brainstorming process. When dealing with industrial design problems, it is frequently difficult to identify the optimal solutions during the initial conceptual design stage. In most cases, a near-optimal solution represents the best solution which can reasonably be hoped for. However, in the automatic design support system proposed in this study, the likelihood of the final two-colour solution closely approximating the optimal solution can be increased by carefully defining the colour-evaluation algorithms used by the genetic algorithm during the search process.

The feasibility and effectiveness of the proposed design support system is demonstrated using the two-colour design of a thermos flask for illustration purposes. The visible, external components of the flask are assumed to be plastic and to have the same glossy surface quality and texture. In other words, the design support system presented in this study focuses solely on the colour-combination aspects of product design. The design support system comprises two subsystems, namely an image prediction system and a colour-combination search system. Briefly, the role of the image prediction system is to evaluate the likely consumer perception of a particular two-colour design, while that of the colour-combination system is to search for the two-colour design which most closely matches the specified consumer perception goal. The proposed design support system is integrated with the commercial I-DEAS system to create a powerful design platform which enables a designer to view 3-D coloured models rendered with specified colour parameters or to search for suitable colour schemes which satisfy the required image evaluation targets.