Palmprint recognition system is very robust and effective biometric technology for accurately identifying a person, based upon physiological characteristics. In the field of biometrics, feature subset selection and classification plays a major role to carry out the identification and authentication process. This paper presents a novel and an accurate palmprint segmentation method to extract the central part of palmprint features based on Region Of Interest (ROI) extraction strategy. A hybrid approach to palmprint recognition system is also proposed in this paper by integrating contourlet transform with the Principal Component Analysis (PCA) which reduces the overall dimensionality and produces accurate results. The Support Vector Machine (SVM) is used as a feature classifier to improve the recognition performance.