A Novel Approach for Quaternion Algebra Based JSEG Color Texture Segmentation

In this work, a novel colour quantization approach has been applied to the JSEG texture segmentation using quaternion algebra. As rule, fundamental vectors of space are derived by inverting three RGB directions in complex hyperplanes. proposed system, is represented as because algebra provides very intuitive means working with homogeneous coordinates. This representation views pixel point three-dimensional space. A that makes use projective geometry and level set methods produced consequence suggested model. The will technique. new utilises binary moment preserving thresholding methodology, therefore splintering clustering method. method used segment clusters found inside cube consistency throughout spectrum both considered. results compared well most recent standard techniques. These comparisons show technique more robust.