The Approach to Image Retrieval Combining Energy Moment and Color Clustering
Ying Deng · Journal of Henan University · 2012
The efficiency and accuracy are important issues of the content-based image retrieval technology.Firstly,an improved soft clustering algorithm for high dimentional data is used in image retrieval.And then,the energy moment based on discrete cosine transform is presented as an efficient feature descriptor.Based on this,a new approach to image retrieval using combining energy moment and color clustering is proposed.The experimental results show that clustering pretreatment can greatly reduce the search scope of the image,thus the accuracy and speed of the retrieval results also improve greatly.And at the same time,the proposed annular energy moment is very robust in terms of scale and rotation invariance.