Chroma based colour enhancement for improved colour segmentation

Abstract

There are many applications rely on the use of colour segmentation in image processing. The major challenge with colour segmentation is that there is no unique solution as each person perceives and segments the colours in images differently. The segmentation results are also heavily influenced by environment factors such as lighting conditions and noise levels. In addition, existing methods tend to focus on the segmentation stage alone and employ little to none for pre-processing. Thus it can be a challenge task to develop an efficient colour segmentation method. In this paper, a novel colour enhancement method is introduced to boost the colour saliency of the critical regions and to improve the consistency of segmentation results by maximizing chroma while preserving hue. Tests in various common colour models using Euclidean distance segmentation method have shown that images pre-processed with the novel colour enhancement method produced much more accurate and reliable colour segmentation results than images without the enhancement or enhancement using existing methods.