Multi view face recognition using multiple camera networks is an active research area. The main aim of this paper is to handle different pose variations in multi camera network and recognizing face from those videos. The traditional approaches handle the pose estimation explicitly ,the proposed work will handle the multiple views of the poses .For a given set of multi view video sequences we use particle filter to track the 3D location of the head. The texture map is generated by back projecting the multi view video. The proposed work is developed using the Spherical Harmonic (SH) representation of the face from the texture mapped on to the sphere. A robust feature is constructed based on the properties of SH projection.