Innovative quaternion algebra-based segmentation for improved jpeg color texture analysis
Bharat Tripathi, Nidhi Srivastava, Amod Kumar Tiwari · Journal of Autonomous Intelligence · 2024
Color texture analysis is a critical component in various computer vision and image processing applications, including object recognition, medical imaging, and remote sensing. Traditional methods like J-Segmentation for color texture segmentation often struggle with capturing complex textures and maintaining color fidelity, especially when dealing with JPEG-compressed images. Quaternion Algebra’s unique ability to represent and manipulate color information in a multidimensional space allows for more accurate feature extraction and segmentation. Our approach Quaternion neural network (QNN) not only improves the segmentation accuracy but also preserves the visual quality of the segmented regions in JPEG images. We demonstrate the effectiveness of our method through extensive experimentation on diverse datasets, showcasing its superiority over existing techniques. The proposed approach not only achieves state-of-the-art results in terms of segmentation accuracy but also offers computational efficiency. This innovation holds great promise for applications in image analysis, computer vision, and medical imaging, where accurate color texture segmentation is paramount.