Image Segmentation using Clustering Algorithm | Hierarchical Agglomerative K-Means

Content Based Image Retrieval (CBIR) is a set of techniques for retrieving semantically relevant images from an image database based on automatically derived image features or image content. From early CBIR, it can be seen that the low level features applied in representing images are often global features which are extracted from an entire image. However the performance of these CBIR approaches is still far away from user’s expectation. The problem can be due to the following two reasons. First, it is not unusual that targets, for which the user searches through an image retrieval system, are not images, but visual objects in images. Global features extracted from the image cannot represent the characteristics of objects in these images. Second, features used in most CBIR works are low-level features (colour, texture and shape etc). The semantic gap between low-level feature and high level semantic understanding of images are often hard to bridge.