Abstract
In this project, Smart Character Recognition System II, the developers have been appreciated the development of character recognition techniques and theirs important. That confidently make us to expect the recognition’s result would be more efficient than the past. Therefore, we have had an idea to integrate many of recognition techniques in order that there would be the best recognition result. There are two stages in the processing. First, find the group of character to reduce the search space of , K-Nearest Neighborhood classifier by using Convolutional Neural Network. In the second stage, K-Nearest Neighborhood is applied by using distances from bounding box to character boundary and Pyramid Histogram of Oriented Gradient as features, recognition result can be improved significantly in this stage.