Wood Image Retrieval Algorithm Based on Keyblock Distribution
Wei Guo Song, Cheng En Cai · 2009
As is known to all, there are many kinds of wood. If we distinguish the wood's characteristics by our eyesight, we can't distinguish the category and property of the wood correctly, and this way will cost enormous workload. In this paper, we propose the keyblock distribution based wood image retrieval algorithm, in order to make the wood image retrieval algorithm more precise and objective. Keyblocks are the representative blocks that we extract from all the candidate blocks. All the keyblocks form the codebook which is used to encode the wood image. By encoding the wood image into the form of index matrix we can extract the feature vector of the wood image. This keyblock distribution based wood image retrieval algorithm is similar to the keyword based text retrieval algorithm. In the text retrieval algorithm, keyword can be used to represent the content of the text. Similarly, the keyblock in our proposed algorithm can be used to represent the content characteristics of wood images. Keyblock can extract the spatial distribution of wood image's characteristics. Experimental results show that our proposed algorithm has achieved ideal performance.