This paper presents a novel approach for classifying leaf based on its five types of venation: palmate, parallel, pinnate, uninervous and reticulate. This novel approach is called Binary Directional Pattern (BiDirP); it outperforms traditional Edge Orientation Histogram (EOH) because it reduces the influence of outlier data or noise if the summation of distance in histogram method is used. Besides that, BiDirP index is closer to human perspective and easily categorizes leaves into their venation state as simply as matching their BiDirP index based on their BiDirP venation range. This method is less time consuming as it does not require powerful classifiers that train for longer time in order to get better performance. Overall, BiDirP outperforms the EOH by 11.48% in accuracy; for BiDirP, percentage accuracy of venation classification is 96.02% while it is 84.54% for EOH.