基于树莓派的文字识别系统设计

ext and images are an important source for human beings to obtain external information, and automatic recognition and extraction of them have important practical significance, this paper focuses on the study of text detection and recognition algorithms, and compares the traditional text detection and recognition algorithms with the text detection and recognition algorithms based on deep learning. Firstly, the MSER+NMS algorithm is used to detect and recognize text in simple scenes. Secondly, based on the CTPN algorithm, as a detection algorithm for complex scenes, by adding the Hough transformation algorithm to correct the oblique text image and the image super-resolution reconstruction algorithm (SRCNN), the input image is preprocessed to ensure that the input image will not interfere with the accuracy of CTPN detection. Again, the CRNN algorithm is used as the algorithm for system text recognition. In order to improve the feature extraction capability of the text recognition network, the CRNN algorithm is improved, and an Inception V3 network is added to the convolutional layer, combining the two networks to improve the performance of the CRNN algorithm to extract features by increasing the depth of the network and building a deeper convolutional network. Finally, with Raspberry Pi as the implementation platform of the system, a text detection and recognition system was designed by using deep learning technology and Raspberry Pi hardware.