Feature selection, optimization and performance analysis of classifiers for biological images

The core objective of this paper is to improve the performance of Content Based Image Retrieval (CBIR) system for biological image by intelligent selection of discriminative feature sets from the set of canonical features. The performance of the CBIR system can be further enhanced by proper selection of Classifier and fine tuning model parameters to obtain improved classification accuracy. We extracted canonical set of features from biological images using a popular tool (WNDCHRM) [3]. We adopted two step approaches for the selection of features. The first step is to partition the canonical feature set into four distinct feature sets. The second step is to perform Principal Component Analysis (PCA) and Fisher Score based selection of features from the partitioned features, applied as training data for different Classifier implementations such as Bayesian and Support Vector Machine (SVM) Classifiers.The performances of Classifiers were analyzed. The results were compared with the results available for classifier. We used IICBU-2008 benchmark biological image data set for our experiments.