A NEW METHOD FOR COLOUR APPEARANCE MEASUREMENT - PARTIAL COLOUR MATCHING: IMPLEMENTATION AND EXPERIMENTATION

摘要

Colour is one of the most critical factors to affect visual appearance. Colour is usually described in terms of unique hues (white, black, red, green, yellow, and blue). Hue scaling and hue cancellation are commonly used methods to measure unique hues. However none of them are able to objectively define unique hues without naming colours. In this paper, we present a new psychophysical method, partial colour matching, to measure unique hues. Observers are presented with a pair of Munsell chips and requested to say whether they have any common shade. We choose 20 Munsell chips, each of which is selected one by one as a reference, while others are classified according to whether they have the common shade with the reference colour. Observation results are organised in a colour wheel, based upon which a mathematical model is developed to derive unique hues objectively without naming colours. This idea is significant. However, there are some difficulties in implementation and experimentation. In this paper, we focus on algorithm and experimental data processing aspects. With regard to the algorithm aspect, we present an algorithm to find chromaticity classes, the core of identifying unique hues. Chromaticity class is defined as the largest set of chips, all of which partially match each other. It is initially time consuming due to chip-to-chip comparison. However, we found some computational redundancy in such comparisons because a chromaticity class may contain repetitive sub classes - matching classes, which denotes chips with exactly the same partial matches. By extracting matching classes, only one chip from each matching class participates in the calculation, which dramatically reduces the computational cost. Moreover, we formalised the algorithm in matrix operations for the easy implementation in Matlab, which further reduce the computing time. Our testing programme showed the algorithm works reliably in real time. In experimental data analysis, we present a new statistical method, homogeneity test, to suppress the unstable boundary observations. Homogeneity test provides us with a way to determine whether it is safe to amalgamate two populations in the observation matrix, combining them together would save the number of repetitions. It is a non-parametric, small-sample-size test. The sum of the lower and upper p-values of the Fisher test is employed as the test statistic. The probability distribution of the statistic under the null (homogeneity) hypothesis is evaluated to obtain corresponding p-values. The test requires only that all observations are statistically independent. In short, we present how we implement colour partial matching a new psychophysics approach to measure colour appearance and the methodologies we used, such as homogeneity test and algorithms. The research shows that: 1) human beings perceive colours by unique hues, which are different to each individuals; 2) Partial colour matching provides us with a method to derive unique hues objectively without naming colours. The methodologies we used are expected not only to lead to the success of partial colour matching but also to benefit other psychophysics research.