Research And Implementation Of The Optical Character Recognition Based On Android Platform

With the rapid development and popularization of the smartphone market, Android operating system holds the highest market proportion. At the same time, people want to enjoy more abundant and comfortable function by smart phone. The combination of Optical Character Recognition and Android devices is useful to the automatic input of information. It is significant to research Optical Character Recognition technology because of the poor ability of processing image on Android platform and the requirement of high reliable Android application. In addition, the process of Optical Character Recognition is relatively fixed.The thesis is focused on pre-processing algorithm of Optical Character Recognition and Optical Character Recognition process. The following two improvements are completed. Firstly, based on the analysis of traditional image processing algorithm, the thesis designs an appropriate pre-processing algorithm for Android platform by adding a step of text image judgment. Secondly, an improved image skew detection method based on directional white run-length is proposed. The improvements include the check range of algorithm and the number of mapping direction, the details are described as follows. Sub-region of text image serves as the input of the algorithm and the number of the original 360 mapping directions is reduced to 180. After that, a skew detection algorithm that aimed to detect and correct the large skew angle of image is implemented by the combination of improved image skew correction algorithm and Android Matrix.And then, integrating all pre-processing steps by Android Native Development Kit technology and Java Native Interface to complete entire pre-processing algorithm. In Optical Character Recognition process, the thesis is focused on character segmentation, feature extraction and classification. And making some improvements in existing character segmentation based on the length of space between characters.In order to verify the effectiveness of the proposed algorithm, a recognition system that is built on the template matching classifier is accomplished. It includes four functional modules, which are image collection, pre-processing, character recognition and system settings. In the realization of the entire system, the image collection module consisted of custom camera and system gallery, pre-processing module uses improvedalgorithm to complete image pre-processing, character recognition module is responsible for recognizing text, system settings module is in charge of image saving path and switch of pre-processing.According to the results of the functional test and performance test of the Optical Character Recognition system, the proposed algorithm could improve the effectiveness of Optical Character Recognition and the interaction with system, which has reached the anticipated goals. It promotes the later research about Optical Character Recognition application.