Handwritten Character Recognition Model Based on Discriminant Convolutional Neural Network

doi:10.3778/j.issn.1002-8331.2208-0108

Abstract

Abstract: At present, the deep learning models for online and offline handwritten character recognition are generally shallow in depth, and they are trained by using softmax loss function, so they lack the ability to learn discriminative information from different classes of samples. In order to build a deeper convolutional neural network model, a convolutional neural network model combining encoder and decoder is proposed, which can effectively increase the network depth and avoid the problem of training difficulties caused by the sharp increase of model parameters. For further improving the recognition accuracy, this paper presents a discriminative loss function. The loss function trains the neural network model by minimizing the distance between the feature of training sample extracted by neural network and the optimized prototype. Since the optimized prototype is learned by the minimization classification error algorithm, the constructed convolutional neural network model in this paper trained with the discriminant loss function can obtain strong discriminative ability. Comparative experiments are carried out on the in-air handwritten Chinese character dataset IAHCC-UCAS2016 and the offline handwritten character dataset MNIST. The results show that compared with softmax loss function, the recognition rate of discriminative loss function is improved by 1.04 percentage points on in-air handwritten dataset. The recognition rates of the proposed discriminant convolution neural network model on the two datasets are 95.53% and 99.73% respectively. Compared with the existing traditional models and depth network models for handwritten character recognition, the proposed model achieves higher recognition accuracy.

Key words: handwritten character recognition, deep learning, softmax loss function, optimized prototype, minimization classification error algorithm

摘要： 目前用于离线和联机手写字符识别的深度学习模型一般深度较浅，且使用softmax损失函数进行训练，缺乏从不同类样本中学习判别信息的能力。为构建更深层卷积神经网络模型，提出了一种编码器与解码器相结合的卷积神经网络模型，该模型可以有效增加网络深度，避免模型参数剧增导致训练困难的问题。为了进一步提高深度学习模型的识别精度，提出了一种判别损失函数，该损失函数通过最小化训练样本在全连接层的输出与该样本同类别优化原型之间的距离来训练神经网络模型。由于优化原型由最小化分类误差算法学习得到，使用判别损失函数训练的卷积神经网络模型具备了更强的判别能力。在公开的联机手写汉字数据集IAHCC-UCAS2016和离线手写字符数据集MNIST上进行对比实验，实验结果表明，在联机手写数据集上与softmax损失函数相比，判别损失函数的识别率提高了1.04个百分点。在两个数据集上，与现有的用于手写字符识别的传统模型和深度模型相比，提出的判别卷积神经网络模型识别率分别达到95.53%和99.73%，获得了更高的识别精度。

关键词: 手写字符识别, 深度学习, softmax损失函数, 优化原型, 最小化分类误差算法

QU Xiwen, WU Xiang, HU Mianjun, HUANG Jun. Handwritten Character Recognition Model Based on Discriminant Convolutional Neural Network[J]. Computer Engineering and Applications, 2023, 59(22): 151-157.

屈喜文, 吴响, 胡冕军, 黄俊. 判别卷积神经网络的手写字符识别模型[J]. 计算机工程与应用, 2023, 59(22): 151-157.

References

[1] ZHANG X Y，BENGIO Y S，LIU C L.Online and offline handwritten Chinese character recognition：a comprehensive study and new benchmark[J].Pattern Recognition，2017，61：348-360.
[2] GAN J，WANG W Q，LU K.A new perspective：recognizing online handwritten Chinese characters via 1-dimensional CNN[J].Information Sciences，2019，478：375-390.
[3] 卓天天，桑庆兵.注意力机制与复合卷积在手写识别中的应用[J].计算机科学与探索，2022，16（4）：888-897.
ZHUO T T，SANG Q B.Application of attention mecha-nism and composite convolution in handwriting recognition[J].Journal of Frontiers of Computer Science and Technology，2022，16（4）：888-897.
[4] QU X W，WANG W Q，LU K，et al.Data augmentation and directional feature maps extraction for in-air handwritten Chinese character recognition based on convolutional neural network[J].Pattern Recognition Letters，2018，111：9-15.
[5] YANG H M，ZHANG X Y，YIN F，et al.Convolutional prototype network for open set recognition[J].IEEE Transactions on Pattern Analysis and Machine Intelligence，2022，44（5）：2358-2370.
[6] BAI Z L，HUO Q.A study on the use of 8-directional features for online handwritten Chinese character recognition[C]//Proceedings of the 2005 International Conference on Document Analysis and Recognition.Washington：IEEE Computer Society，2005：262-266.
[7] YANG W X，JIN L W，TAO D C，et al.DropSample：a new training method to enhance deep convolutional neural networks for large-scale unconstrained handwritten Chinese character recognition[J].Pattern Recognition，2016，58：190-203.
[8] LAI S X，JIN L W，YANG W X.Toward high-performance online HCCR：a CNN approach with DropDistortion，path signature and spatial stochastic max-pooling[J].Pattern Recognition Letters，2017，89：60-66.
[9] GAN J，WANG W Q，LU K.Compressing the CNN architecture for in-air handwritten Chinese character recognition[J].Pattern Recognition Letters，2020，129：190-197.
[10] REN H Q，WANG W Q，LU D，et al.An end-to-end recognizer for in-air handwritten Chinese characters based on a new recurrent neural networks[C]//Proceedings of the 2017 IEEE International Conference on Multimedia and Expo.Washington：IEEE Computer Society，2017：841-846.
[11] REN H Q，WANG W Q，LIU C L.Recognizing online handwritten Chinese characters using RNNs with new computing architectures[J].Pattern Recognition，2019，93：179-192.
[12] LIU C L，NAKAGAWA M.Evaluation of prototype learning algorithms for nearest-neighbor classifier in application to handwritten character recognition[J].Pattern Recognition，2001，34：601-615.
[13] BAI Z L，HUO Q.A study of nonlinear shape normalization for online hand-written Chinese character recognition：dot density vs.line density equalization[C]//Proceedings of the 18th International Conference on Pattern Recognition.Washington：IEEE Computer Society，2006：921-924.
[14] IOFFE S，SZEGEDY C.Batch normalization：accelerating deep network training by reducing internal covariate shift[C]//Proceedings of the 32nd International Conference on Machine Learning，2015：448-456.
[15] MAAS A L，HANNUN A，NG A.Rectifier nonlinearities improve neural network acoustic models[C]//Proceedings of the 30th International Conference on Machine Learning，2013：16-21.
[16] LIN M，CHEN Q，YAN S C.Network in network[EB/OL].（2013-12-16）[2022-06-10].https：//arxiv.org/abs/1312.4400.
[17] HE K M，ZHANG X，REN S Q，et al.Deep residual learning for image recognition[C]//Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition，2016：770-778.
[18] SRIVASTAVA N，HINTON G，KRIZHEVSKY A，et al.Dropout：a simple way to prevent neural networks from overfitting[J].The Journal of Machine Learning Research，2014，15（1）：1929-1958.
[19] LECUN Y，BOTTOU L，BENGIO Y，et al.Gradient-based learning applied to document recognition[J].Proceeding of the IEEE，1998，86（11）：2278-2324.
[20] KINGMA D，BA J.Adam：a method for stochastic optimization[C]//Proceedings of the 3rd International Conference for Learning Representations，San Diego，2015.
[21] CAI S，ZHANG L，ZUO W，et al.A probabilistic collaborative representation based approach for pattern classification[C]//Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition，2016：2950-2959.
[22] XU J，AN W P，ZHANG L，et al.Sparse，collaborative，or nonnegative representation：which helps pattern classification?[J].Pattern Recognition，2019，88：679-688.
[23] CIREGAN D C，MEIER U，GAMBARDELLA L M，et al.Convolutional neural network committees for handwritten character classification[C]//Proceedings of the 12th International Conference on Document Analysis and Recognition，Beijing，2011：1250-1254.
[24] CIREGAN D C，MEIER U，MASCI J，et al.Flexible，high performance convolutional neural networks for image classification[C]//Proceedings of the 2011 International Joint Conference on Artificial Intelligence，2011：1237-1242.
[25] KIMURA K，TAKASHINA K，TSURUOKA S，et al.Modified quadratic discriminant functions and its application to Chinese character recognition[J].IEEE Transactions on Pattern Analysis and Machine Intelligence，1987，9（1）：149-153.
[26] QU X W，WANG W Q，LU K，et al.In-air handwritten Chinese character recognition with locality-sensitive sparse representation toward optimized prototype classifier[J].Pattern Recognition，2018，78：4783-4797.
[27] XU N，QU X W，WANG W Q.A discriminative classifier for in-air handwritten Chinese characters recognition[C]//Proceedings of the 7th International Conference on Internet Multimedia Computing and Service，Zhangjiajie，2015：16-19.