Recognizing Natural Scene Characters by Convolutional Neural Network and Bimodal Image Enhancement

Zhu, Yun; Sun, Jun; Naoi, Satoshi

doi:10.1007/978-3-642-29364-1_6

Cited by 14 publications

(10 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…3 shows the results of our method compared with those of [6,7,1,2]. The overall RER of our system reaches 9.03% compared to 14.04% of the best method [2] and it ranges from 32.14% for seriously distorted images to 5.03% for clear images. Compared with the other methods, the performance of our system behaves better especially in cases such as multi-color, uneven light, little contrast and blurring.…”

Section: Comparison With Existing Methodsmentioning

confidence: 74%

See 1 more Smart Citation

Image character recognition using deep convolutional neural network learned from different languages

Bai

Chen

Feng

et al. 2014

2014 IEEE International Conference on Image Processing (ICIP)

View full text Add to dashboard Cite

This paper proposes a shared-hidden-layer deep convolutional neural network (SHL-CNN) for image character recognition. In SHL-CNN, the hidden layers are made common across characters from different languages, performing a universal feature extraction process that aims at learning common character traits existed in different languages such as strokes, while the final softmax layer is made language dependent, trained based on characters from the destination language only. This paper is the first attempt to introduce the SHL-CNN framework to image character recognition. Under the SHL-CNN framework, we discuss several issues including architecture of the network, training of the network, from which a suitable SHL-CNN model for image character recognition is empirically learned. The effectiveness of the learned SHL-CNN is verified on both English and Chinese image character recognition tasks, showing that the SHL-CNN can reduce recognition errors by 16-30% relatively compared with models trained by characters of only one language using conventional CNN, and by 35.7% relatively compared with state-of-the-art methods. In addition, the shared hidden layers learned are also useful for unseen image character recognition tasks.

show abstract

Section: Comparison With Existing Methodsmentioning

confidence: 74%

“…On a same test set with [6,7], Saidane's method got better result. Zhu [2] first find an optimum projection from color to grayscale conversion, and then using CNN to do scene character recognition. Jacobs [3] also proposed a CNN based character recognition method.…”

Section: Related Workmentioning

confidence: 99%

Image character recognition using deep convolutional neural network learned from different languages

Bai

Chen

Feng

et al. 2014

2014 IEEE International Conference on Image Processing (ICIP)

View full text Add to dashboard Cite

show abstract

“…Gabor filters Local subspace classifier with appearancebased features scheme for recognition of characters in natural images [19] and mixture of HOG-based features for contextual spatial information extraction [20] are among the alternative methods presented. Recently, works such as that by Zhu et al [21] reported a convolutional neural network-based scene character or text recognition to be more competitive. Extensive review of the literature on segmented character recognition reveals that methods relying entirely on global image feature detection and description are not employed for segmented character recognition in natural scene images.…”

Section: Segmented Character Recognitionmentioning

confidence: 99%

Segmented character recognition using curvature-based global image feature

Chekol¹,

Çelebi²,

Tasci³

2019

Turk J Elec Eng & Comp Sci

View full text Add to dashboard Cite

Character recognition in natural scene images is a fundamental prerequisite for many text-based image analysis tasks. Generally, local image features are employed widely to recognize characters segmented from natural scene images. In this paper, a curvature-based global image feature and description for segmented character recognition is proposed. This feature is entirely dependent on the curvature information of the image pixels. The proposed feature is employed for segmented character recognition using Chars74k dataset and ICDAR 2003 character recognition dataset. From the two datasets, 1068 and 540 images of characters, respectively, are randomly chosen and 573-dimensional feature vector is synthesized per image. Quadratic, linear and cubic support vector machines are trained to examine the performance of the proposed feature. The proposed global feature and two well-known local feature descriptors called scale invariant feature transform (SIFT) and histogram of oriented gradients (HOG) are compared in terms of classification accuracy, computation time, classifier prediction and training time. Experimental results indicate that the proposed feature yielded higher classification accuracy (%65.3) than SIFT (%53), performed better than HOG and SIFT in terms of classifier training time, and achieved better prediction speed than HOG and less computational time than SIFT.

show abstract

“…Doküman içinde tespit edilen figürler ve başlıkları MongoDB'de tutularak kullanıcı etkileşimli sorgulama/arama imkânı sunulmuştur. Son zamanlarda araştırmacılar karmaşık arkaplana ait görüntülerden metin saptama ile ilgili çalışmalar yapmışlardır [13][14]. Takip eden bölümde mimari detaylı olarak açıklanacaktır.…”

Section: İlgi̇li̇ çAlişmalarunclassified

DocDig: Content Based Figure Search in Digitized Documents

Eken

Atay

Sönmez

et al. 2018

Düzce Üniversitesi Bilim Ve Teknoloji Dergisi

View full text Add to dashboard Cite

ÖZET Örüntü tanıma psikolojiden biyometriye, biyoenformatikten gen ifadelerinin analizine, trafikten hesaplamalı finansa kadar birçok alanda kullanılmaktadır. Optik Karakter Tanıma da bu alanlardan bir tanesidir. Kamu ve özel birçok firma, arşivlerindeki klasörlenmiş verilerini taratarak dijital hale getirmekte ve bunun için emek yoğun çalışmalar yapmaktadır. Ancak resim olarak sayısallaştırılan bu verilerin içerik olarak aranması ve işlenmesi ancak operatörlerin manuel olarak taranan resim verisine meta veri eklemesi ile kısmi olarak gerçekleşmektedir. Bu çalışmada, resim olarak taranan ve sayısal hale getirilen büyük miktarlardaki bu dokümanlar üzerinde içerik bazlı figür aramalarını mümkün kılan bir mimari geliştirdik. Kullanıcı, bazı anahtar kelimelerle arama yaparak sayısal dökümanlardaki ilgili figürleri başlıklarıyla beraber görüntüleyebilmektedir. Sistemin yapılabilirlik ve başarımı farklı veri setleri üzerinde test edilmiş, başarılı sonuçlar elde edilmiştir. ABSTRACTPattern recognition is used in many areas, from psychology to biometrics, analysis of gene expressions from bioinformatics, from traffic to finance calculated. Optical Character Recognition is also one of these areas. Many public and private firms digitize their archived data and make labor-intensive studies for this purpose. However, the retrieval and processing of these data, which are digitized as images, is only partially realized by adding metadata to the manually scanned image data. In this work, we developed an architecture that makes contentbased figure searches possible on these scanned documents in large quantities. The user can search with some keywords and display related figures in digital documents with their captions. The feasibility and performance of the system have been tested on different data sets and successful results have been obtained.

show abstract

Recognizing Natural Scene Characters by Convolutional Neural Network and Bimodal Image Enhancement

Cited by 14 publications

References 10 publications

Image character recognition using deep convolutional neural network learned from different languages

Image character recognition using deep convolutional neural network learned from different languages

Segmented character recognition using curvature-based global image feature

DocDig: Content Based Figure Search in Digitized Documents

Contact Info

Product

Resources

About