The Application of Sub-Pattern Approach in 2D Shape Recognition and Retrieval

In 2D shape recognition and retrieval approach using shape feature extraction, statistical shape analysis methods such as PCA, ICA and NMF are commonplace, and these methods using subspace approach, have not been adequately investigated for recognition and retrieval of 2D shapes. The main hurdle in achieving higher recognition efficiency seems to be the shape sensitivity. In this paper we suggest, subspace method approach. The main idea is to use modular technique to improve the recognition and retrieval efficiency. Normally in the earlier methods proposed so far, a complete image is considered in training and matching process, in modular method approach partial image is used for training and matching the 2D images. The recognition and retrieval process is carried out in two phase, in the first phase uses the ridgelet transform applied. The second phase PCA is used for dimensionality reduction and to extract the effective features. For recognition and retrieval a study was conducted by using seventeen different distance measure technique. The training and testing process is conducted using leave-one-out strategy. The retrieval process is carried out by considering standard test “bull eyes” score. The proposed method is tested on the standard dataset MPEG-7. The experiment results of Subspace ridgelet PCA are compared with Subspace PCA method.