Community Research and Development Information Service - CORDIS

Development of a new index for colour description

Interpretation of the significance of coloration patterns and colour changes in fish is complicated by the lack of data on skin chromaticity parameters, and by difficulties in expressing colour differences in a way that correlate more with our understanding of the three dimensional nature of colour. The measurement of chromaticity parameters provides a means of quantifying colour. Based on the opponent colours theory of vision, the CIELAB scales provide single values (L, a, b) that can be used to express the red (+a) or green (-a), and yellow (+b) or blue (-b). Comparisons between specimens can be easily expressed as colour differences in terms of £GL, £Ga, and £Gb.

A total colour difference can also be calculated from theses parameters, but cannot distinguish whether the differences are lightness (£GL) changes only or more severe hue changes or differences in saturation of a colour. Therefore, an alternative method for the expression of colour differences that correlate more with our basic understanding of the three dimensional nature of colour has been developed (CIE £GL £GC £GH) that has added the concept of hue-angle (hab).

Hue is determined by the dominant wavelength and is the name of a colour as found in its pure state in the spectrum. Thus, hue is the attribute of colour perception by which an object is judged to be red (0„a), yellow (90„a), green (180„a), blue (270„a), and so forth. Chroma refers to the intensity or purity of hue, regardless of how light or dark it is, ranges from neutral to brilliant, and is a measure of how much gray and white light is mixed in with the ¡¥pure¡¦ focal colour.

However, for comparisons between different specimens the mean hue has to be calculated. Most of the published literature on fish skin or fillet colour performs hue comparison between different samples with the classical statistical approach, without asking whether there is, in fact, a mean direction among the population of data that were sampled. Therefore, for testing significance of the mean hue and for multisample testing of hue the nature of the variable, being from circular distributions should be considered.

Data from circular distributions may not be analysed using conventional statistics for reasons stemming from the arbitrariness of the zero point on the circular scale. Statistical methods for describing and analysing data from circular distributions are relatively new and are still undergoing development. These methods were applied in the present work both for descriptive statistics and hypothesis testing for the hue variable. This approach offered also the possibility to develop a new parameter, the Entire Colour Index, which by definition is a combined variable of Hue and Chroma, and provide a good approximation of the actual colour.

Reported by

University of Crete
714 09 Heraklion
See on map
Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top