In this paper we propose a biometric sclera recognition and validation system. Here the sclera segmentation is performed by a time-adaptive active contour-based region growing technique. The sclera vessels are not prominent so image enhancement is required and hence a bank of 2D decomposition Haar wavelet multi-resolution filters is used to enhance the vessels pattern for better accuracy. For feature extraction, Dense Scale Invariant Feature Transform (D-SIFT) is used. D-SIFT patch descriptors of each training image are used to form bag of features by using k-means clustering and a spatial pyramid model, which is used to produce the training model. Support Vector Machines (SVMs) are used for classification. The UBIRIS version 1 dataset is used here for experimentation. An encouraging Equal Error Rate (EER) of 0.66% is attained in the experiments presented. Keywords: Biometric; Sclera vessel patterns; D-SIFT; SVM; Bag of features, k-means, Bank of 2D decomposition Haar multi- resolution filters wavelet. .