You are on page 1of 3

Topic of the Project/Title (Domain: Image Processing) Content based image retrieval using clustering.

Objective and Scope Content-based image retrieval has widespread applications in image processing. The motivation behind content-based image retrieval is for users to avoid annotation of images. In some systems text based annotation and retrieval of images is inadequate. It might include images from surveillance cameras, finger-print recognition or x-ray pictures. Motivated by the lack of an efficient image retrieval technique, content-based image retrieval was introduced. Content-based means that the technology makes direct use of content of the image rather than relying on human annotation of metadata with keywords.

Process Description This project is about retrieving a collection of images from the image database that is similar to the query image. The user passes a query image as the input, processing will happen and the output will be a collection of images similar to the query image. Once the query image is given as input, features will be extracted from the image which is stored in an excel file. On these features any of the three clustering algorithm (k-means clustering, hierarchical clustering and contribution-based clustering) is applied. Finally the output will be a collection of images similar to the query image. Finally a comparison of the three clustering algorithms (k-means clustering, hierarchical clustering and contribution-based clustering) should be carried out.

Content-based Image Retrieval(CBIR) System.

Images

Clustering Algorithm

Image Clusters

Query Image

Similarity Comparison

Retrieved Images

The steps are as follows: Query image is passed to the system Calculate feature vectors of the query image as well as the images in the database using Discrete Cosine Transform Group the feature vectors based on the similarity by applying K-means clustering algorithm Calculate the similarity difference between query image and the images in database Retrieve the similar images Group the feature vectors based on the similarity by applying contribution-based clustering algorithm Calculate the similarity difference between query image and the images in database Retrieve the similar images

Group the feature vectors based on the similarity by applying contribution-based clustering algorithm Calculate the similarity difference between query image and the images in database Retrieve the similar images Compare the performance of the three different clustering algorithms

Resources Software used: Matlab7.0 or higher Conclusion Finally a comparison of the three clustering algorithms are made and their performance is compared.

You might also like