Typical single-chip CCD colour camera provides output signal containing RGB components of acquired image. Strong inter-channel correlation of that signal, unacceptable for some applications, is shown. Therefore, some kind of RGB colour space transform is required to decrease this correlation. Only the Karhunen-Loève transform (KLT) secures complete decorrelation. However, possible applications of this method are limited by the fact that coefficients of transform matrix must be calculated for each image separately. In the paper details of effective implementation such transform of colour images are described. The examples of KLT applications in the field of colour image processing are quoted. Using the standard colour images, principal properties of KLT are shown, i.e. a complete decorrelation of colour components and energy compaction in first components. This paper presents also an approach that allows determining KLT coefficients (new KLT approximation) that are constant for certain class of images and guarantee complete decorrelation simultaneously. A methodology for finding these coefficients is briefly described and some remarks on criteria of image selection for this class are made.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.