With the tremendous development in Networking and Multimedia technologies, Image Retrieval plays significant roleand is used for browsing, searching and retrieving images from a large database of digital images. Image Retrieval techniques utilize annotation methods of adding metadata such as captioning, keywords or descriptions to the images. The manual image annotation is much time consuming laborious and expensive.As the data bases size increases, annotation becomes a tedious task. Thus automatic image annotationhas drawn the attention of the researchers in recent years.. The increase in social web application and the semantic web drawn attention of researchers in the development of several web-based image retrieval tools. This paper presents an easy, efficient image retrieval approach using a new image feature descriptor called Micro-Structure Descriptor (MSD). The microstructures are defined based on an edge orientation similarly while the MSD is built based on the underlying colors in micro-structures with similar edge orientation.In this method of Image retrieval the MSD extracts features by simulating human’s early visual processing by effectively integrating color, texture, color lay out information and shape. The proposed MSD algorithm has high indexing performance and low dimensionalityas it has only 72 dimensions for full color images.The technique is examined on Corel datasets with natural images; the results demonstrate that this image retrieval method is much more efficient and effective than reprehensive feature descriptors, such as Gabor features and Multi Texton Histograms.
[1]
Marc Alexa,et al.
An evaluation of descriptors for large-scale image retrieval from sketched feature lines
,
2010,
Comput. Graph..
[2]
Benoit Huet,et al.
Shape recognition from large image libraries by inexact graph matching
,
1999,
Pattern Recognit. Lett..
[3]
Qionghai Dai,et al.
Efficient view-based 3-D object retrieval via hypergraph learning
,
2014,
Tsinghua Science and Technology.
[4]
Thierry Pun,et al.
Performance evaluation in content-based image retrieval: overview and proposals
,
2001,
Pattern Recognit. Lett..
[5]
Martin Vetterli,et al.
Shapes From Pixels
,
2015,
IEEE Transactions on Image Processing.