Convolution neural network based text image classifications

• Main Author: Zhou, Xiang
• Other Authors: Yu Hao
• Format: Final Year Project
• Language: English
• Published: 2017
• Subjects: DRNTU::Engineering::Electrical and electronic engineering
• Online Access: http://hdl.handle.net/10356/71667
• Institution: Nanyang Technological University

Convolution neural network(CNN) is a sensor with multiple layers, which is designed for identifying 2-dimensional images, with parallel processing ability, self-learning ability and good fault tolerance. In dealing with 2-dimensional graphics problems, especially for the identification of misplacement, zooming and other distortion invariant’s forms applications, it has a good robustness and operational efficiency, and it has been widely used in various types of image recognition This paper introduces its model principle and specific approaches, as well as its application in image classification, namely traffic sign identification and handwritten number recognition. CNN combines the extracting features and identification process for training the neural network, and has achieved great success in the field of image classification. The experimental part of this paper uses CNN models for traffic sign and handwritten number recognition, and the correct rate is superior to other traditional methods.