Deformation of iris pattern caused by pupil dilation and contraction is one of the most influential intra-class variations. Most state-of-the-art iris recognition methods only focus on the description of local iris texture features. We believe that both geometric and photometric features are important to achieve a robust matching result of deformed iris images. This paper proposes to decompose iris images into lowpass and bandpass components using nonsubsampled contourlet transform (NSCT) and then extract different features. Geometric features are extracted in bandpass components based on key point detection to align deformed iris patterns. And then aligned Ordinal features are extracted in lowpass components to characterize the ordinal measures of local iris regions. Finally, key point features in bandpass components and Ordinal features in lowpass components are fused for deformed iris image matching. Extensive experiments on two challenging iris image databases namely CASIA-Iris-Lamp and ICE'2005 demonstrate that the proposed method outperforms state-of-the-art methods in deformed iris recognition.