Illuminant and device invariant colour using histogram equalisation

Abstract

Colour can potentially provide useful information for a variety of computer vision tasks such as image segmentation, image retrieval, object recognition and tracking. However, for it to be helpful in practice, colour must relate directly to the intrinsic properties of the imaged objects and be independent of imaging conditions such as scene illumination and the imaging device. To this end many invariant colour representations have been proposed in the literature. Unfortunately, recent work (Second Workshop on Content-based Multimedia Indexing) has shown that none of them provides good enough practical performance.In this paper we propose a new colour invariant image representation based on an existing grey-scale image enhancement technique: histogram equalisation. We show that provided the rank ordering of sensor responses are preserved across a change in imaging conditions (lighting or device) a histogram equalisation of each channel of a colour image renders it invariant to these conditions. We set out theoretical conditions under which rank ordering of sensor responses is preserved and we present empirical evidence which demonstrates that rank ordering is maintained in practice for a wide range of illuminants and imaging devices. Finally, we apply the method to an image indexing application and show that the method out performs all previous invariant representations, giving close to perfect illumination invariance and very good performance across a change in device.

Real-time illumination invariant lane detection for lane departure warning system

“…where o and c represent different illumination conditions and a; b and c are diagonal coefficients which indicate the relationship between o and c illumination conditions (Finlayson, Hordley, Schaefer, & Tian, 2005). The rank order is preserved under the change of illumination and following relation can be derived as follows:…”

Combining color and spatial information for object recognition across illumination changes

“…The mean percentage of such pairs of corresponding pixels reaches 55%. We have shown, in a previous work, that another assumption which stipulates that the rank measures of corresponding pixels are preserved across illumination changes (Finlayson et al, 2005) has been only verified for 6% of corresponding pixels in the three similar color images of Fig. 1 (Muselet et al, 2005).…”

“…Funt has tested several invariant features on the database of the Simon Fraser University (SFU) and has demonstrated that the object recognition results obtained by the intersection between the invariant histograms processed by the greyworld normalization outperforms those obtained by the intersection between classical invariant histograms (Funt et al, 1998). Furthermore, Finlayson have used the East Anglia University (EAU) database and has concluded that the intersection between the invariant histograms resulting from one-dimensional histogram equalization provides better results than those obtained by greyworld normalization (Finlayson et al, 2003). Hence, we propose to use these two databases and to compare the results obtained by these two schemes with those obtained by the mean intersection between the adapted co-occurrence matrices.…”

“…Other transformations require an analysis of the color distribution. The greyworld normalization of the color component levels (Gershon et al, 1988), the moments based on the rotation of the color distribution (Lenz et al, 1999), the principal component analysis of the color distribution (Lin et al, 2002) and the equalization of the color histogram (Finlayson et al, 2005) yield functions with which the color histograms are transformed into invariant histograms.…”

“…Instead of considering directly the color vectors of pixels to justify complex illumination change assumptions, as those used by Finlayson et al (1995), Gevers and Smeulders (1999), Gershon et al (1988), Lenz et al (1999), Lin et al (2002), and Finlayson et al (2005), we introduce the spatial rank measures of the color component levels, a concept with interesting properties in case of illumination changes.…”