We have worked on the configuration management plan report in this week. In addition, I have researched about the image processing part. The k-means clustering algorithm is commonly used in computer vision as a form of image segmentation. The results of the segmentation are used to aid border detection and object recognition. We can use this algorithm to detect object for feature extraction. However, the standard Euclidean distance is insufficient in forming the clusters in our case due to space dimension histogram. Therefore, Mahalanobis distance can be used instead of Euclidean distance to form clusters better in large histogram space dimensions. ISODATA algorithm is a good modified version of k-mean clustering algorithm. It uses both an iterative version of native k-mean clustering algorithm to find the best number of clusters (K) and Mahalanobis distance.
Reference:
Fatos T. Yarman-Vural, Ergin Ataman: Noise, histogram and cluster validity for Gaussian-mixtured data. Pattern Recognition 20(4): 385-401 (1987)
No comments:
Post a Comment