作者机构：
[Li, Zhi; Sun, Yaqi; Xie, Xiaolan] Guangxi Normal Univ, Sch Comp Sci & Engn, Guilin, Guangxi, Peoples R China.;[Xie, Xiaolan] Guilin Univ Technol, Sch Informat Sci & Engn, Guilin, Guangxi, Peoples R China.;[Yang, Kai; Sun, Yaqi] Hengyang Normal Univ, Sch Comp Sci & Technol, Hengyang, Peoples R China.
通讯机构：
[Xie, XL ] G;Guangxi Normal Univ, Sch Comp Sci & Engn, Guilin, Guangxi, Peoples R China.;Guilin Univ Technol, Sch Informat Sci & Engn, Guilin, Guangxi, Peoples R China.
关键词：
Computer vision;Super-resolution;Scene text image;Batch-transformer;Loss function
摘要：
Recognizing low-resolution text images is challenging as they often lose their detailed information, leading to poor recognition accuracy. Moreover, the traditional methods, based on deep convolutional neural networks (CNNs), are not effective enough for some low-resolution text images with dense characters. In this paper, a novel CNN-based batch-transformer network for scene text image super-resolution (BT-STISR) method is proposed to address this problem. In order to obtain the text information for text reconstruction, a pre-trained text prior module is employed to extract text information. Then a novel two pipeline batch-transformer-based module is proposed, leveraging self-attention and global attention mechanisms to exert the guidance of text prior to the text reconstruction process. Experimental study on a benchmark dataset TextZoom shows that the proposed method BT-STISR achieves the best state-of-the-art performance in terms of structural similarity (SSIM) and peak signal-to-noise ratio (PSNR) metrics compared to some latest methods.
会议名称：
2015 16th International Conference on Electronic Packaging Technology (ICEPT)
关键词：
Principal Component analysis;Surface Mount Technology;Character recognition
摘要：
In this paper, we present an approach to recognizing characters in surface mount technology (SMT) product. Firstly, we capture the SMT characters image by a camera. Some feature information which can represent the SMT product character images is extracted. Secondly, by processing and analyzing the feature information, Principal Component analysis is used in characters recognition. Lastly, a SMT characters recognition system based on VC++ is developed. The experiment has shown that the system has some advantages in acquisition speed, readability and portability, and practical.
摘要：
In this paper, a compressing and reconstruction method for a noise video based on Compressed Sensing (CS) theory is proposed. At first, the CS theory is presented. Then the noise video is estimated from noisy measurement by solving the convex minimization problem. The video recovery algorithms based on gradient-based method is used to compressing and reconstructing the noise signal. And a compressive sensing algorithm with gradient-based method is proposed. At last, the performance of the proposed approach is shown and compared with some conventional algorithms. Our method can obtain best results in terms of peak signal noise ratio (PSNR) than those achieved by common methods with only a little runtime.
作者机构：
College of Electrical and Information Engineering, Hunan University, Hunan;410082, China;Department of Computer Science, Hengyang Normal University, Hengyang;421008, China;[赵辉煌; 王耀南] College of Electrical and Information Engineering, Hunan University, Hunan
通讯机构：
[Zhao, H.] C;College of Electrical and Information Engineering, Hunan University, Hunan, China
