Chinese Characters Recognition from Screen-Rendered Images Using Inception Deep Learning Architecture

Xu, Xin; Zhou, Jun; Zhang, Hong; Fu, Xiaowei

doi:10.1007/978-3-319-77380-3_69

Cited by 6 publications

(12 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The memory required by this method is large, and the accuracy of the variable length recognition cannot be guaranteed. In order to obtain the character-level recognition accuracy, Xu et al [10] adopted the method of character segmentation and designed a novel character recognition model. Although the accuracy can be obtained based on character segmentation, the threshold needs to be adjusted in each detection picture.…”

Section: Screen Rendered Text Detectionmentioning

confidence: 99%

“…In the character segmentation part, this paper uses vertical projection method to over-segment each character, and then character fusion correction (refer to [10]) is employed for rectifying over-segmentation. Character fusion correction compares each over-segmented word width with the mean of the whole image.…”

Section: ) Character Segmentationmentioning

confidence: 99%

“…Due to the addition of the Gaussian removal module, the removal of extreme values degrade the recognition performance. To alleviate this problem, in the character recognition phase, this paper proposes a new character recognition network model after modifying previous work [10]. The specific network model diagram is shown in Figure 2.…”

Section: B Character Recognitionmentioning

confidence: 99%

“…After that, Connectionist Text Proposal Network (CTPN) [12] is used, which further improves the detection performance of Chinese complex text. In order to maintain high recognition accuracy, the character segmentation based on the word width fusion correction [10] is performed in this paper. Because the word width of special FIGURE 1.…”

Section: Introductionmentioning

confidence: 99%

“…In order to avoid removing below average values, the recognition network with better performance is employed, for the character recognition stage. Using Inception-v4 replace Inception-v2 in [10] work as shown in Figure 1 dotted box (3). In the character recognition stage, the character recognition model proposed in this paper is evaluated both by the self-made dataset and the ICDAR (International Conference on Document Analysis and Recognition) 2013 dataset.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

A Novel Method of Detecting Chinese Rendered Text on Tilted Screen of Mobile Devices

Zheng

Wang

2020

IEEE Access

Self Cite

View full text Add to dashboard Cite

Due to the development of the Internet, screen rendered text detection and recognition have wide applications such as screen capture translation, medical text reading, screen text assisted reading and so on. A large number of screen-rendered attentions have been paid to the English text detection, multi-language scene detection and indefinite length recognition. However, these studies mainly focus on getting the screen rendered image from the device itself. In the actual scene, lots of screen rendered text is mainly captured by the mobile device, and text image has a certain capturing angle. Because of the tilted text line, the current method has a severe degradation in detection performance. This paper proposes a Chinese-style scene text detection method for tilting screens capturing by mobile devices. Firstly, in order to obtain robust detection performance, the Connectionist Text Proposal Network (CTPN) method is employed for horizontal text detection. Secondly, to overcome the dependence of detection on horizontal text, the lightweight Oxford Visual Geometry Group network (VGGnet) is used to correct the tilted direction of the whole image. Moreover, the word width of special characters and punctuation characters are far below average value, so this paper uses the Gaussian removal method to delete the narrow word width and designed a more accurate network to recognize characters. Experiments show that based on the comprehensive scheme, this paper achieves excellent results, even in some extreme bad cases. INDEX TERMS Tilted text detection, mobile application, character segmentation, optical character recognition.

show abstract

Section: Screen Rendered Text Detectionmentioning

confidence: 99%

Section: ) Character Segmentationmentioning

confidence: 99%

Section: B Character Recognitionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations