Detection of Global Salient Region via High Dimensional Color Transform and Local Spatial Support

This paper proposes novice automatic salient region detection in an image which includes both the global and local features. The main motivation behind this approach is to construct a saliency map by utilizing a linear combination of colors in a high dimensional color space. In general, the human perception is highly complicated and non-linear and in response to that, the salient region consists of distinct colors compared to the background. The estimation of an optimal construction of a saliency map was done by agglomerating the low-dimensional colors to the high-dimensional feature vectors. Furthermore, a relative location and color contrast between super pixels are utilized to improve the performance. It was tested under three distinct datasets to evaluate the applicability and practicability of our proposed method. Citation: T.Jeyapriya, Mepco Schlenk Engineering College; G.Rajasekaran ,Mepco Schlenk Engineering College. "Detection of Global Salient Region via High Dimensional Color Transform and Local Spatial Support." Global Research and Development Journal For Engineering : 126 - 132.