Classification of document blocks using density feature and connectivity histogram

doi:10.1016/0167-8655(95)00039-J

Pattern Recognition Letters

Volume 16, Issue 9, September 1995, Pages 955-962

https://doi.org/10.1016/0167-8655(95)00039-J Get rights and content

Abstract

In this paper, we present a document block classification algorithm to automatically classify different types of blocks embedded in a document image. Two kinds of features, density feature and connectivity histogram, are devised to achieve the classification goal. In our approach, segmented document blocks are first classified into text and non-text blocks via the density feature. Then, the connectivity histogram is utilized to further classify non-text blocks into image and graphics blocks. Experimental results reveal the feasibility of the new technique in classifying document blocks.

References (8)

K.C. Fan et al.
Segmentation and classification of mixed text/graphics/image documents
Pattern Recognition Lett.
(1994)
T. Pavlidis et al.
Page segmentation and classification
CVGIP: Graphical Models and Image Processing
(1992)
D. Wang et al.
Classification of newspaper image blocks using texture analysis
Computer Vision, Graphics, and Image Processing
(1989)
G. Nagy et al.
Hierarchical representation of optically scanned documents

There are more references available in the full text version of this article.

Cited by (17)

An automatic document processing system for medical data extraction
2015, Measurement: Journal of the International Measurement Confederation
Citation Excerpt :
Indeed, confusing one exam for another can be dangerous in this kind of application. As regards the general aspects of the processing flow, it should be noted that the extraction and classification of data requires a lot of automatic decisions about where data are located and what is their meaning, and each decision may have deep consequences on subsequent processing [35]. This is in contrast with other industrial applications, where the authors have experienced less correlation between subsequent processing steps [37–41].
This paper illustrates an automatic document processing system for the extraction of data contained in medical laboratory results printed on paper. The final goal of the research is to automate the collection of medical data and to enable an efficient management and dissemination of the information. The following processing steps of the system are described in detail: image preprocessing; layout analysis for the identification of the tables contained in the document; extraction and classification of the laboratory results. Among the many features of the system there are the use of an open source OCR engine, as a basis of further processing, and the storage in XML format of the data retrieved, for ease of sharing. The knowledge base used to guide the data extraction is also explained. The proposed approach has been tested on several document formats and performance analyzed.
Parameter-free based two-stage method for binarizing degraded document images
2012, Pattern Recognition
Citation Excerpt :
In the past decades, many document image operations, such as document layout analyzing [1,2], data hiding [3], skew estimation [4,5], stroke extraction [6], document block classification [7], and optical character recognition [8], have been developed for storing, transmitting, and managing digital documents.
Binarization plays an important role in document image processing, especially in degraded documents. For degraded document images, adaptive binarization methods often incorporate local information to determine the binarization threshold for each individual pixel in the document image. We propose a two-stage parameter-free window-based method to binarize the degraded document images. In the first stage, an incremental scheme is used to determine a proper window size beyond which no substantial increase in the local variation of pixel intensities is observed. In the second stage, based on the determined window size, a noise-suppressing scheme delivers the final binarized image by contrasting two binarized images which are produced by two adaptive thresholding schemes which incorporate the local mean gray and gradient values. Empirical results demonstrate that the proposed method is competitive when compared to the existing adaptive binarization methods and achieves better performance in precision, accuracy, and F-measure.
Document zone content classification and its performance evaluation
2006, Pattern Recognition
Citation Excerpt :
Based on the description of white space inside regions, it classified a given region into text, graphics, line art regions. Fan and Wang [24] presented a document block classification algorithm using density feature and connectivity histogram. The attribute of each segmented block is divided into three classes: text, graphics, and image.
This paper describes an algorithm for the determination of zone content type of a given zone within a document image. We take a statistical based approach and represent each zone with 25 dimensional feature vectors. An optimized decision tree classifier is used to classify each zone into one of nine zone content classes. A performance evaluation protocol is proposed. The training and testing data sets include a total of 24,177 zones from the University of Washington English Document Image database III. The algorithm accuracy is 98.45% with a mean false alarm rate of 0.50%.
Skew detection and block classification of printed documents
2001, Image and Vision Computing
Since the number of daily-received paper-based office documents is overwhelming, the development of document image analysis, which converts the paper-based documents into electronic forms becomes increasingly important. This paper describes a skew detection method which first smoothes the black runs and locates the black–white transitions to emphasize the text lines. Then the skew angle is determined by an improved Hough transform. For the block classification step, a rule-based classifier is presented. The classification rules are derived from the gray level entropy, block aspect ratio, and run length analysis. To evaluate the performance of the proposed methods, a test set of 100 different documents is used. The results of the experiments reveal that all of the 100 documents are successfully skew-corrected and the precision rate and the recall rate of the proposed block classifier are satisfactory.
Adaptive page segmentation for color technical journals' cover images
1998, Image and Vision Computing
Page segmentation to locate text blocks is a prior and primary step in document processing, in particular for understanding a journal's cover page. However, texts, graphics and images are usually isolated in most documents, unlike cover pages in which texts may be overlaid onto graphics or images. In this paper a new adaptive page segmentation method is proposed to extract text blocks from various types of color technical journals' cover images. Although color involves useful information to overcome the overlapping problem, color processing requires tremendous computation loads. Thus, a complexity analysis is included to adaptively adjust processing steps in our approach. In other words, simple cover images, with few colors and no text–graphics/image overlapping, can be treated as monochrome images to speed up processing time, while for complex cover images, with many colors and text–graphics/image overlapping, correct segmentation results can still be obtained but more processing time is required. To accomplish the design concept mentioned above, our method includes several components. First, in order to degrade the processing complexity on true color images, a new simple quantization method is employed to reduce the color numbers from 24-bit true colors to 42 colors or less. In the block segmentation stage, smearing, labeling and complexity analysis techniques are used together with edge and color information to find out coherent blocks adaptively. After that, in the block classification stage, some conventional and some new features are computed from each block to decide whether it is a text block or not. Finally, in the post-processing stage, some spatial relations are adopted to rectify the classification results. Experimental results prove the feasibility and practicality of the proposed approach.
Reading order of Chinese newspaper articles using a block-growing method
1998, Image and Vision Computing
The reading order of articles plays an important role in the areas of document analysis and document understanding. Since the reading sequence conveys significant semantic information embedded in a document, it will influence the robustness and correctness of post-applications in document analysis. Hence, the reading order of documents, which most researchers and systems frequently ignore, is indispensable for post-processing in order to understand document systems. In this paper, a block-growing approach is proposed to generate the reading order of Chinese newspapers, by matching predefined style graphs. In this graph-matching approach, the geometric relation graph (GRG) is first constructed according to the geometric relationships among the segmented blocks of input documents. Those blocks that belong to the same article are gradually merged to form an article by matching the predefined style graphs. Since the local reading information is retained during the block-growing process, the global reading information can be easily generated. The proposed bottomup merging approach is powerful and flexible in finding the reading order of Chinese newspaper articles. A wide variety of Chinese newspapers with horizontal and vertical styles were tested to verify the validity of our proposed method. Experimental results reveal the feasibility and effectiveness of our proposed approach.

View all citing articles on Scopus

^☆: This work is supported by National Science Council of Taiwan under grant NSC-83-0408-E-008-001.

View full text

Classification of document blocks using density feature and connectivity histogram☆

Abstract

Pattern Recognition Lett.

CVGIP: Graphical Models and Image Processing

Computer Vision, Graphics, and Image Processing

Hierarchical representation of optically scanned documents