Segmentation of Terahertz imaging using k-means clustering based on ranked set sampling

摘要: A novel approach of segmentation Terahertz images is proposed.We explain how to reformulate the k-means technique under ranked set sample.The goal estimate good centers and classify data.We compare our based on simple random sampling technique. imaging a modality that has been used with great potential in many applications. Due its specific properties, this type makes possible discrimination diverse regions within sample. Among methods, clustering considered as one most popular techniques. However, it known especially sensitive initial starting centers. In paper, we propose an original version for images, called ranked-k-means, which essentially less initialization We present design sample expected well observed data. Our tested various real images. Experimental results show more efficient than other techniques such fundamental technique, standard Bradley refinement