In this paper, we experimentally evaluate the validities of dimension reduction for image pattern recognition. Biometric such as face recognition, iris recognition, and finger print recognition are achieved as image pattern recognition. Image pattern recognition uses pattern recognition techniques for classification of image data. For the numerical achievement of image pattern recognition techniques, images are sampled using the array of pixels. This sampling procedure derives vectors in higher dimensional metric space from image patterns. For accurate achievement of pattern recognition techniques, the dimension reduction of data vectors are essential methodology, since time and space complexities of data processing depends on the dimension of data. However, dimension reduction causes information loss of geometrical and topological features of image patterns. Desired dimension reduction selects appropriate low-dimensional subspace preserving the information for classification. We experimentally evaluate effects of the dimension reduction techniques which are used as preprocessing of pattern recognition of image data.