Text Localization Based on Fast Feature Pyramids and Multi-Resolution Maximally Stable Extremal Regions

Zamberletti, Alessandro; Noce, Lucia; Gallo, Ignazio

doi:10.1007/978-3-319-16631-5_7

Cited by 40 publications

(10 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore, the mainstream of text detection methods always focused on the structure of individual characters and the relationships between characters [40], e.g. connected component based methods [38,27,22]. These methods often use stroke width transform (SWT) [9] or maximally stable extremal region (MSER) [20,24] to first extract character candidates, and using a series of subsequence steps to eliminate non-text noise for exactly connecting the candidates.…”

Section: Related Workmentioning

confidence: 99%

Deep Matching Prior Network: Toward Tighter Multi-oriented Text Detection

Liu

Jin

2017

2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)

312

171

View full text Add to dashboard Cite

Detecting incidental scene text is a challenging task because of multi-orientation, perspective distortion, and variation of text size, color and scale. Retrospective research has only focused on using rectangular bounding box or horizontal sliding window to localize text, which may result in redundant background noise, unnecessary overlap or even information loss. To address these issues, we propose a new Convolutional Neural Networks (CNNs) based method, named Deep Matching Prior Network (DMPNet), to detect text with tighter quadrangle. First, we use quadrilateral sliding windows in several specific intermediate convolutional layers to roughly recall the text with higher overlapping area and then a shared Monte-Carlo method is proposed for fast and accurate computing of the polygonal areas. After that, we designed a sequential protocol for relative regression which can exactly predict text with compact quadrangle. Moreover, a auxiliary smooth Ln loss is also proposed for further regressing the position of text, which has better overall performance than L2 loss and smooth L1 loss in terms of robustness and stability. The effectiveness of our approach is evaluated on a public word-level, multioriented scene text database, ICDAR 2015 Robust Reading Competition Challenge 4 "Incidental scene text localization". The performance of our method is evaluated by using F-measure and found to be 70.64%, outperforming the existing state-of-the-art method with F-measure 63.76%.

show abstract

Section: Related Workmentioning

confidence: 99%

Deep Matching Prior Network: Toward Tighter Multi-oriented Text Detection

Liu

Jin

2017

2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)

312

171

View full text Add to dashboard Cite

show abstract

“…Precision Recall F-measure Neumann and Matas [18] 73 65 69 Shi et al [22] 83 63 72 Bai et al [3] 79 68 73 Zamberletti et al [32] 86 70 77 Tian et al [23] 85 76 80 Zhang et al [33] 88 74 80 Our model 88 72 79 Table 1: Localization performances on ICDAR2013(%).…”

Section: Resultsmentioning

confidence: 99%

Detecting Arbitrary Oriented Text in the Wild with a Visual Attention Model

Huang

Yang

et al. 2016

Proceedings of the 24th ACM International Conference on Multimedia

View full text Add to dashboard Cite

Text embedded in images provides important semantic information about a scene and its content. Detecting text in an unconstrained environment is a challenging task because of the many fonts, sizes, backgrounds, and alignments of the characters. We present a novel attention model for detecting arbitrary oriented and curved scene text. Inspired by the attention mechanisms in the human visual system, our model utilizes a spatial glimpse network to processes the attended area and deploys a recurrent neural network that aggregates the information over time to determine the attention movement. Combining this with an off-the-shelf region proposal method, the model achieves the state-of-the-art performance on the highly cited ICDAR2013 dataset, and the MSRA-TD500 dataset which contains arbitrary oriented text.

show abstract

“…The overall performance of the proposed algorithm is evaluated, and compared with other main MSER-based text detection algorithms in recent years in Table III. [7] 0.75 0.83 0.79 Neumann et al [9] 0.65 0.74 0.69 Zamberletti et al [10] 0.71 0.80 0.75…”

Section: Experiments Results and Analysismentioning

confidence: 99%

Natural Scene Text Detection Based on MSER

Wang¹,

Li²,

Liu³

et al. 2018

Proceedings of the 2018 3rd International Conference on Communications, Information Management and Network Security (CIMNS 2018

View full text Add to dashboard Cite

Scene text detection has important applications in the fields of intelligent transportation, industrial automation, multimedia retrieval and so on. This paper employs the improved MSER algorithm combined with convolutional neural network for scene text detection. Gradient amplitude enhancement processing is used to enhance the text boundary before a combinational suppression strategy is applied to filter out coincident regions, approximately coincident regions and nested regions. Then the Char-CNN classifier is designed to classify the candidate regions. A hierarchical clustering algorithm is used to merge the candidate regions, and finally generate the text position information. We evaluate the algorithm performance on the ICDAR2013 dataset. The results show that the improved MSER algorithm increases the recall rate of the character region from 88.9% to 90.2%, and the proportion of character regions in the candidate regions increases from 3.25% to 35.19%. And the classification accuracy of Char-CNN is 93.6%. The recall and accuracy rate of the algorithm are 0.68 and 0.85 respectively, and the F-Measure value is 0.76. Compared with existing scene text detection algorithms, the proposed algorithm has a competitive overall performance.

show abstract

Text Localization Based on Fast Feature Pyramids and Multi-Resolution Maximally Stable Extremal Regions

Cited by 40 publications

References 20 publications

Deep Matching Prior Network: Toward Tighter Multi-oriented Text Detection

Deep Matching Prior Network: Toward Tighter Multi-oriented Text Detection

Detecting Arbitrary Oriented Text in the Wild with a Visual Attention Model

Natural Scene Text Detection Based on MSER

Contact Info

Product

Resources

About