A new efficient binarization method: application to degraded historical document images

Hadjadj, Zineb; Cheriet, Mohamed; Meziane, Abdelkrim; Cherfa, Yazid

doi:10.1007/s11760-017-1070-2

Cited by 24 publications

(10 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We have carried out a detailed comparative analysis of our proposed method against seven notable methods, namely Otsu's [15], Niblack's [12], Sauvola's [22], Rivest-Hénault's [21], Hadjadj's [6], Hadjadj's [5], and Hadjadj's [7] methods. Otsu's [15] method is based on global thresholding technique; whereas (c) Niblack's [12] (d) Sauvola's [22] (e) Rivest-Hénault's [21] (f) Hadjadj's [6] (g) Hadjadj's [5] (h) Hadjadj's [7] (i) Proposed method Niblack's [12] and Sauvola's [22] methods are adaptive thresholding based. All the other remaining methods are active contour model based methods.…”

Section: Comparison With State-of-the-art Methodsmentioning

confidence: 99%

“…Few works have also been reported on document image binarization using active contour models [5][6][7]21]. In [7], Chan-Vese active contour model with Sauvola's binarization method [22] is used for binarization of document images with various object sizes.…”

Section: Introductionmentioning

confidence: 99%

“…Document images with complex background and large variation in background intensities can not be binarized efficiently with this method. Also, methods [5,6,21] have been tested on very few datasets. To show the efficiency of the proposed work, we have compared our work with previously mentioned methods [5][6][7]21].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Degraded Document Image Binarization Using Active Contour Model

Gupta

Bag

2021

Communications in Computer and Information Science

View full text Add to dashboard Cite

Document image binarization separates the foreground from background which is a very crucial pre-processing step in OCR. Accuracy of binarization immensely influences the accuracy of OCR. Various degradations like inadequate illumination, complex background, ink bleed, smear, etc. make binarization a challenging exercise. In this paper, we propose a novel method for binarization based on Active Contour Model (ACM) which is different from current thresholding techniques. Instead of calculating a global or local threshold value, it uses the energyminimization concept of ACM to accomplish binarization. Background of the document image is estimated adaptively and then removed to handle noises in degraded documents. Active contour model is then applied to the image to get the binary output. Comprehensive experimentation has been completed effectively with benchmark DIBCO series datasets. A comparison with other existing methods verifies the efficiency of the proposed method.

show abstract

Section: Comparison With State-of-the-art Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Degraded Document Image Binarization Using Active Contour Model

Gupta

Bag

2021

Communications in Computer and Information Science

View full text Add to dashboard Cite

show abstract

“…Other techniques that were introduced later, such as Niblack [13], Sauvola [14], and [15], depend on the estimation of the local intensity using local threshold-based techniques. The binary image resulting from the previous techniques might consider selected background pixels as it is applied locally in each pixel neighborhood [16]. To overcome this issue, Lu et al [17] suggested fusing the local and global thresholds and began by applying polynomial smoothing to classify the background and subsequently used the local threshold to capture the foreground text.…”

Section: A First Categorymentioning

confidence: 99%

Binarization of Degraded Document Images Using Convolutional Neural Networks and Wavelet-Based Multichannel Images

2020

View full text Add to dashboard Cite

Convolutional neural networks (CNNs) have previously been broadly utilized to binarize document images. These methods have problems when faced with degraded historical documents. This paper proposes the utilization of CNNs to identify foreground pixels using novel input-generated multichannel images. To create the images, the original source image is decomposed into wavelet subbands. Then, the original image is approximated by each subband separately, and finally, the multichannel image is constituted by arranging the original source image (grayscale image) as the first channel and the approximated image by each subband as the remaining channels. To achieve the best results, two scenarios are considered, that is, two-channel and four-channel images, and then fed into two types of CNN architectures, namely, single and multiple streams. To investigate the effect of the multichannel images proposed as network inputs, the CNNs used in the architectures are three popular networks, namely, Unet, SegNet, and DeepLabv3+. The experimental results of the scenarios demonstrate that our method is more successful than the three CNNs when trained by the original source images and proves competitive performance in comparison with state-of-the-art results using the DIBCO database.

show abstract

“…Therefore, numerous local adaptive methods have been proposed for preserving the textual information as accurately as possible. Many experiments were carried out for investigating and examining the available local adaptive binarization methods like [51][52][53][54][55][56].…”

Section: Binarizationmentioning

confidence: 99%

Deep Learning Algorithms for Arabic Handwriting Recognition: A Review

Al-Saffar¹,

Awang²,

Al-Saiagh³

et al. 2018

IJET

View full text Add to dashboard Cite

Computer vision (CV) refers to the study of the computer simulation of human visual science. Major task of CV is to collect images (or video) so that they could be used for analysis, gathering information, and making decisions or judgements. CV has greatly progressed and developed in the past few decades. In recent years, deep learning (DL) approaches have won several contests in pattern recognition and machine learning. (DL) dramatically improved the state-of-the-art in visual object recognition, object detection, handwritten recognition and many other domains. Handwritten recognition technique is one of this tasks that targeted to extract the text from documents or another images type. In contrast to the English domain, there are a limited works on the Arabic language that achieved satisfactory results, Due to the Arabic language cursive nature that induces many technical difficulties. This paper highlighted the pre-processing and binarization methods that have been used in the literature along with proposed numerous directions for developing. We review the various current deep learning approaches and tools used for Arabic handwritten recognition (AHWR), identified challenges along this line of this research, and gives several recommendations including a framework based (DL) that is particularly applicable for dealing with cursive nature languages.

show abstract

A new efficient binarization method: application to degraded historical document images

Cited by 24 publications

References 18 publications

Degraded Document Image Binarization Using Active Contour Model

Degraded Document Image Binarization Using Active Contour Model

Binarization of Degraded Document Images Using Convolutional Neural Networks and Wavelet-Based Multichannel Images

Deep Learning Algorithms for Arabic Handwriting Recognition: A Review

Contact Info

Product

Resources

About