I am doing work in content based image retrieval in that I am using clustering techniques.
If you want to cluster images based on colour (or intensity) you can use vector quantization type of clustering e.g. K-mean clustering. There are quite a few approaches where colour coherence is used for clustering similar coloured pixels; or clustering based on colour-spatial information where along with the colour, position of the pixel also plays role in clustering.
Then there are also texture based clustering approaches which will keep texture of the pixel (and neighbourhood) in mind while selecting the cluster.
Which feature/s do you have in mind to use for the CBIR?
Please look into the following, I hope this will help you in CBIR
Article Content Based Image Retrieval Using Combined Features of Sha...
There are five techniques namely k-means, ISODATA, merging, splitting and mean shift techniques
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering
- Signal Processing
- Image Processing