Holistic handwritten word recognition considers the whole input image as a single word, indivisible unit and recognize the word based on overall shape. In this paper, we present a holistic approach for the off line handwritten word recognition using Gaussian filter with different divergence like horizontal, vertical and 2 diagonal difference to plot the endpoint. The endpoint images are subblocked with different sizes. In each subblock enumerate the number of appearance of dots for feature extraction. The extracted features are fed to SVM classifier to recognize the given holistic word. The proposed system achieves a good recognition accuracy of 3×6 subblock with 24 classes holistic word obtained 90.52%.