A survey of document image classification: problem statement, classifier architecture and performance evaluation

Chen, Nawei; Blostein, Dorothea

doi:10.1007/s10032-006-0020-2

Cited by 145 publications

(76 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…It comprises a set of simple techniques and procedures, which are used to work upon the images of documents and exchange them from pixel information into a format that can be read by a computer. Information retrieval from the document images is a challenging task, hence number of techniques and procedures are used for document image processing [3]. Converting a scanned grey scale image into a binary image, the foreground (or regions of interest) and removing the background is an important step in many image analysis systems includes document image processing.…”

Section: Introductionmentioning

confidence: 99%

A Survey on Document Image Analysis

Vijayarani¹

2018

IJRASET

View full text Add to dashboard Cite

Nowadays document images are becoming very popular and it is using in digitalized libraries and organization. Paper documents can be converted into digital form by using digitization equipment's. Digitalized document images are more compact to store in the computer memory and it becomes inexpensive. These document images also sent and receive via email, fax and other web usages and it simply view and print by users. Information in these document images are more structured and presented in a natural language with the help of a grammar and a script. This paper discusses document image analysis, applications of document images, challenges and issues for handling document images.

show abstract

Section: Introductionmentioning

confidence: 99%

A Survey on Document Image Analysis

Vijayarani¹

2018

IJRASET

View full text Add to dashboard Cite

show abstract

“…SIFT [25] including both the detector and descriptor as a whole system provides better results compared to other systems under Mikolajczyk`s evaluation framework [26] but It's known that SIFT suffers from a high computational complexity [27]. PCA-SIFT is one of the most successful extensions [28], which applies Principal Components Analysis (PCA) [29], a standard technique for dimensional reduction, on the target image patch and produces a similar descriptor as the standard SIFT one. Another algorithm called i-SIFT, has been presented based on the PCA-SIFT by Fritz et al [30], which applied Information Theoretic Selection to cut down the number of interest points so as to reduce the cost for computing descriptors.…”

Section: Introductionmentioning

confidence: 99%

Scale Invariant Feature Transform Plus Hue Feature

B.Daneshvar

2017

Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

ABSTRACT:This paper presents an enhanced method for extracting invariant features from images based on Scale Invariant Feature Transform (SIFT). Although SIFT features are invariant to image scale and rotation, additive noise, and changes in illumination but we think this algorithm suffers from excess keypoints. Besides, by adding the hue feature, which is extracted from combination of hue and illumination values in HSI colour space version of the target image, the proposed algorithm can speed up the matching phase. Therefore, we proposed the Scale Invariant Feature Transform plus Hue (SIFTH) that can remove the excess keypoints based on their Euclidean distances and adding hue to feature vector to speed up the matching process which is the aim of feature extraction. In this paper we use the difference of hue features and the Mean Square Error (MSE) of orientation histograms to find the most similar keypoint to the under processing keypoint. The keypoint matching method can identify correct keypoint among clutter and occlusion robustly while achieving real-time performance and it will result a similarity factor of two keypoints. Moreover removing excess keypoint by SIFTH algorithm helps the matching algorithm to achieve this goal.

show abstract

“…This article is a contribution to the extensive literature on document classification, which is surveyed by Nawei Chen and Dorothea Blostein (2007), especially efforts (such as ours) that make use of image features for categorizing documents. The types of features we use here are by now standard.…”

Section: Introductionmentioning

confidence: 99%

Automatically Determining Versions of Scholarly Articles

Rothchild

Shieber

2017

SRC

View full text Add to dashboard Cite

Background: Repositories of scholarly articles should provide authoritative information about the materials they distribute and should distribute those materials in keeping with pertinent laws. To do so, it is important to have accurate information about the versions of articles in a collection.Analysis: This article presents a simple statistical model to classify articles as author manuscripts or versions of record, with parameters trained on a collection of articles that have been hand-annotated for version. The algorithm achieves about 94 percent accuracy on average (cross-validated).Conclusion and implications: The average pairwise annotator agreement among a group of experts was 94 percent, showing that the method developed in this article displays performance competitive with human experts.

show abstract

A survey of document image classification: problem statement, classifier architecture and performance evaluation

Cited by 145 publications

References 43 publications

A Survey on Document Image Analysis

A Survey on Document Image Analysis

Scale Invariant Feature Transform Plus Hue Feature

Automatically Determining Versions of Scholarly Articles

Contact Info

Product

Resources

About