Textline information extraction from grayscale camera-captured document images

Abstract

Cameras offer flexible document imaging, but with uneven shading and non-planar page shape. Therefore cameracaptured documents need to go through dewarping before being processed by traditional text recognition methods. Curled textline detection is an important step of dewarping. Previous approaches of curled textline detection use binarization as a pre-processing step, which can negatively affect the detection results under uneven shading. Furthermore, these approaches are sensitive to high degrees of curl and estimate x-line 1 and baseline pairs using regression which may result in inaccurate estimation. We introduce a novel curled textline detection approach for grayscale document images. First, the textline structure is enhanced by using match filter bank smoothing and then central lines of textlines are detected using ridges. Then, x-line and baseline pairs are estimated by adapting active contours (snakes) over ridges. Unlike other approaches, our approach does not use binarization and applies directly on grayscale images. We achieved 91% of detection accuracy with good estimation of x-line and baseline pairs on the dataset of CBDAR 2007 document image dewarping contest.

Text Extraction in Document Images: Highlight on Using Corner Points

“…Edge-based methods were mainly used for color images or camera captured images [26] [27][28] [29] but not only [30] [31]. Some were also used as well with connected components analysis [32].…”

Performance evaluation of curled textline segmentation algorithms on CBDAR 2007 dewarping contest dataset

“…The extended coupled-snakelets version 1 contains an additional step of removing badly deformed snakelets pairs on the basis of neighboring snakelets properties. [5,6] In this approach multi-scale and multi-orientated anisotropic Gaussian smoothing is used for enhancing curled textline structure. Then central lines of textlines are detected by using Horn-Riley [16,17] based ridges detection technique.…”

“…In this paper, we describe our publicly available CBDAR 2007 dewarping contest dataset [8] and vectorial performance evaluation metrics [9] for comparing curled textline segmentation algorithms. Here, we have selected following previously reported curled textline segmentation algorithms for performance evaluation and comparison: i) neighborhood-distance based approach [2], ii) baby-snakes [3], iii) coupled-snakelets [4], iv) smoothing and ridges based approach [5,6] and v) skew detection based approach [7]. In future, researchers can use the above mentioned dataset and performance evaluation metrics and can compare their curled textline segmentation results to our baseline.…”

“…There are variety of curled textline segmentation techniques in the literature [2,3,4,5,6,7], but no work has been done for the performance evaluation and comparison of these techniques on common dataset and performance evaluation metrics. In this paper, we describe our publicly available CBDAR 2007 dewarping contest dataset [8] and vectorial performance evaluation metrics [9] for comparing curled textline segmentation algorithms.…”

Text Extraction from Document Images- A Review

“…Bukhari et al [14] proposed a method for curled textline detection from gray-scale camera captured document images. The gray-scale textline was promoted by using multi-oriented multi-scale anisotropic Gaussian smoothing.…”