Table 1 - uploaded by Soumen Bag
Content may be subject to copyright.
Percentage of distortion error of methods proposed by DattaÀParui, Huang et al., and Telea et al. Note that the blank entries in the table indicate very negligible error which we can consider as 0% error.
Source publication
In this paper we propose a thinning methodology applicable to character images. It is novel in terms of its ability to adapt to local character shape while constructing the thinned skeleton. Our method does not produce many of the distortions in the character shapes which normally result from the use of existing thinning algorithms. The proposed th...
Similar publications
The 3D-Printing technology is ready to produce parts with specific properties like individual stiffness. Based on a predefined outer shape, the inner structure of a printed part defines mainly the mechanical features. By Hamilton-Jacobi skeletonization, a stiffness enhancement of a printable shape can be achieved in a way, that a novel AM-corner in...
The thinning methodology is novel in terms of its ability to incorporate character shape specific knowledge while constructing the thinned skeleton. But removal of spurious strokes or shape deformation in thinning is a difficult problem. In this paper, we have proposed a novel medial-axis based thinning strategy used for performing skeletonization...
Citations
... In order to extract the structural features from the segmented conjunct characters, the segmented characters are needed to be thinned. We have utilised the measures namely, junction point distortion and end point distortion, as defined by Bag and Harit [17] to substantiate that the final thinned characters are much less distorted. ...
Substantial size of convoluted conjunct characters in Bengali language makes the recognition process burdensome. In this paper, we propose a structural disintegration based segmentation technique that fragments the conjunct characters into discernible shapes for better recognition accuracy. We use a set of structure based segmentation rules that bifurcates the characters into discernible shape components. The bifurcation is done by finding the touching region where two basic shapes coincide to form a conjunct character. The proposed method has been tested on a data set of Bengali handwritten conjunct characters efficiently. In future, we will continue our work to incorporate it as a prominent preprocessing step for Bengali optical character recognition system.
... Then, fault enhancement is presented for improving the discontinuity resolution and moreover thinning the thick lineaments, and various approaches have been developed and applied to seismic fault detection (e.g., Pedersen et al., 2004;Cohen et al., 2006;Bag & Harit, 2011;Hale, 2013;Wang et al., 2014;Zhang et al., 2014).Among them, the ant tracking (Pedersen et al., 2004) and the lineament thinning (Hale, 2013) are considered most popular with wide applications in the industry. In particular, the former applies the principles of Complimentary Contributor Copy swarm intelligence and describes the collective behavior of ants finding the shortest path between the nest and a food source by communicating via pheromone, a chemical substance that attracts other ants (Pedersen et al., 2004).However, while enhancing the faults in seismic data, such tracking also exaggerates artifacts with weak waveform/amplitude variations, such as processing effects, channel boundaries, and chaotic responses, all of which are magnified to the same level as the faults of interest and could be extracted as fault patches by mistake. ...
... The lineament thinning works by searching for the one-pixel-thick fault surface based on the discontinuity magnitude, so that the contrast between major faults and subtle ones is well preserved without exaggerating non-fault artifacts. Various algorithms have been developed (e.g., Cohen et al., 2006;Bag & Harit, 2011;Hale, 2013;Zhang et al., 2014a), and the scanning one by Hale (2013) is considered most useful in generating the thinned fault surfaces without generating undesired bifurcation branches in the presence of noises and complex boundaries. ...
Process monitoring and early fault detection are paramount in modern industries. A well-developed process monitoring system is expected to reduce failure rates, production costs, and prevent health and environmental hazards. An abundance of sophisticated monitoring and fault detection strategies is available, most of which require either accurate process models, including models of the most frequently occurring faults, or large databases filled with relevant historical data. Unfortunately, neither of the previous is usually available, and process monitoring is often reduced to more-or-less elaborate limit-checking. Even in the case when the models and/or historical datasets are available, the obtained monitoring techniques are highly specialized and must usually be re-developed for any new application at hand. This chapter presents a general-purpose, selfadaptive expert system for online and offline process monitoring with possibility of fault detection, capable to learn the properties of the process being monitored and to identify and separate its working regimes, both nominal and unwanted ones. The proposed system is also capable to identify changes in the process operation; both abrupt and incipient ones. The flexibility of the system is achieved by separating the monitoring algorithm into two distinct layers. The first layer, which is application dependent, is called Feature Generator (FG). FG uses current measurements of process variables to compute “features," numerical values used to concisely describe current working conditions. Features are then fed to the second layer, Potential Based Recursive Clustering (PBRC) algorithm, an original, novel adaptive clustering procedure. In essence, PBRC is designed as self-learning and auto-growing neural-network which tracks, in a robust manner, the motion of the process in the features space. It identifies, in an adaptive fashion, areas corresponding to separate process working conditions and decides which of the working conditions is currently active. If a novel working regime is encountered, PBRC can enlarge its knowledge base and incorporate the novelty for further reference. The system was tested and validated on a number of laboratory installations and also in real industrial environment. Obtained results showed that proposed system can detect faults and changes in working regime in the early stages with high accuracy and reliability.
... Then, fault enhancement is presented for improving the discontinuity resolution and moreover thinning the thick lineaments, and various approaches have been developed and applied to seismic fault detection (e.g., Pedersen et al., 2004;Cohen et al., 2006;Bag & Harit, 2011;Hale, 2013;Wang et al., 2014;Zhang et al., 2014).Among them, the ant tracking (Pedersen et al., 2004) and the lineament thinning (Hale, 2013) are considered most popular with wide applications in the industry. In particular, the former applies the principles of Haibin Di and Dengliang Gao 56 swarm intelligence and describes the collective behavior of ants finding the shortest path between the nest and a food source by communicating via pheromone, a chemical substance that attracts other ants (Pedersen et al., 2004).However, while enhancing the faults in seismic data, such tracking also exaggerates artifacts with weak waveform/amplitude variations, such as processing effects, channel boundaries, and chaotic responses, all of which are magnified to the same level as the faults of interest and could be extracted as fault patches by mistake. ...
... The lineament thinning works by searching for the one-pixel-thick fault surface based on the discontinuity magnitude, so that the contrast between major faults and subtle ones is well preserved without exaggerating non-fault artifacts. Various algorithms have been developed (e.g., Cohen et al., 2006;Bag & Harit, 2011;Hale, 2013;Zhang et al., 2014a), and the scanning one by Hale (2013) is considered most useful in generating the thinned fault surfaces without generating undesired bifurcation branches in the presence of noises and complex boundaries. ...
In petroleum exploration and production, faults are of importance by controlling pathways of hydrocarbon migration and accumulation in the subsurface, and robust fault detection is one of the major tasks of geologic and geophysical interpretation from three-dimensional (3D) seismic data. Traditionally, faults are interpreted by manually picking on vertical/horizontal seismic sections with geologic consistence; however, such manual tool is time consuming and sensitive to the visibility and interpreters’ bias, especially for a large dataset with structural complexities. With the development of new signal processing and data visualization technologies, computer-aided semiautomatic/automatic fault extraction has been the focus of recent geophysical research on seismic fault detection with superiorities in both computational efficiency and result accuracy. Various methods have been developed and implemented in the petroleum industry; however, their results are often limited by the detection resolution and contaminated by non-fault artifacts that interfere with reliable structural interpretation. This paper provides a comparative analysis and comprehensive review of the existing fault detection methods, including manual picking and computer-aided automatic/semi-automatic extraction. A suite of algorithms for each method are tested regarding their computational efficiency and result accuracy by calibrating their results to the original seismic images for quality control. Finally, for addressing the limitations and problems observed in the existing techniques, innovative solutions are presented to further enhance the reliability of fault interpretation from seismic data.
... But if it is a single touching modifier, then check whether the touching points of present and next modifiers are nearer to each other or not. If they are not, then go to step 3. Otherwise, join them by applying the extrapolation method used in [1] and then move to the next step. All different possibilities of upper modifier are shown in Fig. 7. 3. Find the column which is four columns preceding from the touching point (see Fig. 8). ...
The proper character level segmentation of printed or handwritten text is an important preprocessing step for optical character recognition (OCR). It is noticed that the languages having cursive nature in writing make the segmentation problem much more complicated. Hindi is one of the well known language in India having this cursive nature in writing style. The main challenge in handwritten character segmenta-tion is to handle the inherent variability in the writing style of different individuals. In this paper, we present an efficient character segmentation method for handwritten Hindi words. Segmentation is performed on the basis of some structural patterns observed in the writing style of this language. The proposed method can cope with high variations in writing style and skewed header lines as input. The method has been tested on our own database for both printed and handwritten words. The average success rate is 96.93 %. The method yields fairly good results for this database comparing with other existing methods. We foresee that the proposed character segmenattion technique can be used as a part of an OCR system for cursive handwritten Hindi language.
... We have also done a quantitative analysis among Datta and Parui, Huang et al., and ourselves. We have considered different measurement techniques used in [1] to quantify different types of shape distortions as reported earlier. We measure the shape distortion at junction or end points as a ratio of the number of distorted junction (end) points and total number of junction (end) points in whole dataset. ...
Binary image thinning has wide applications in image processing, machine vision, and pattern recognition. Thinning is a preprocessing step to obtain single-pixel-thin skeleton for document imaging and pattern analysis. Indian languages are complex in character shape than Latin, Chinese, Japanese, and Korean languages. The performance of existing thinning methods cannot reach the state of the art in the field of document image processing for Indian languages. Our objective is to improve the performance of complex-shaped character skeletonization, particularly for Oriya script. This paper presents an iterative parallel thinning method which is well suited to handwritten Oriya characters. Some fundamental requirements of thinning like preservation of stroke connectivity, removal of spurious strokes have been ensured in this method. Experimental results show its efficacy comparing with other existing thinning methods.
... The thinned skeleton is required to preserve the structural property of the characters. Producing good quality thinned skeletons of handwritten character images still continue to be a challenge in this field [1] due to the variability in writing style, spilling of ink on poor quality paper and usage of pens which produce thick strokes. Due to these reasons thinning is subject to several distortions like spurious strokes, spurious loops, recession of stroke length, and shape distortion at junction points. ...
Thinning which is an important preprocessing step for character recognition is often subject to several kinds of distortion. Junction point distortion is a major imperfection in thinned images especially for handwritten Indian scripts due to the presence of large number of complicated junctions in them. Such distortion does allow the optical character recognition (OCR) systems to exploit the properties of these junctions for character recognition. We present a novel methodology to reduce distortion at junction regions adjoining the matra for BangIa script. Our method uses geometric properties of the junctions to solve the problem. We have tested our approach on our own data set consisting of a variety of isolated handwritten character images by different writers and have got promising results.
... Thinning is done by removing pixels from the contour of the character either sequentially or in parallel. Although this is the most common type of thinning method, yet there are some other methods of thinning which appears in the literature [2] [3]. Any thinning algorithm must retain the significant features of the pattern, ignoring local noise without introducing distortions of its own. ...
Thinning is an important preprocessing operation
used in different document image processing and analysis
applications. The main objective of thinning is to obtain singlepixel
thin skeleton without any shape distortion. It is noticed
that documents written in ink-sketch pens and scanned with
high precision scanners suffer from high degree of unevenness
on their outer surfaces. This unevenness results in distortions
in the shapes of thinned images. These distortions are mainly
two types namely, spurious loops and spurious strokes. Our
proposed algorithm gets rid of these distortions in the thinned
image. We have tested our approach on our own data set of
about 1500 characters and have got promising results.
The aim of this work is to develop a graph theoretical computer vision framework to partition shape of an image object into parts based on a heuristic approach such that the partitioning remains consistent with human perception. The proposed framework employs a special polygonal approximation scheme to represent a shape suitably in
simpler graph form where each polygonal side represents a graph-edge.
The shape-representative graph is explored to determine vertex-visibility
graph by a simple algorithm presented in this paper. We have proposed
a heuristic based iterative clique extraction strategy to decompose the
shape-representative graph depending on its vertex-visibility graph. This
proposed framework considers MPEG-7 shape data set for probing the
acceptability of the proposed framework and according to our observation, the performance of the framework is comparable with existing
schemes.
The proper character level segmentation of printed or handwritten text is an important preprocessing step for optical character recognition (OCR). It is noticed that the languages having cursive nature in writing make the segmentation problem much more complicated. Hindi is one of the well known language in India having this cursive nature in writing style. The main challenge in handwritten character segmentation is to handle the inherent variability in the writing style of different individuals. In this paper, we present an efficient character segmentation method for handwritten Hindi words. Segmentation is performed on the basis of some structural patterns observed in the writing style of this language. The proposed method can cope with high variations in writing style and skewed header lines as input. The method has been tested on our own database for both printed and handwritten words. The average success rate is 96.93 %. The method yields fairly good results for this database comparing with other existing methods. We foresee that the proposed character segmenattion technique can be used as a part of an OCR system for cursive handwritten Hindi language.
Thinning of character images is a big challenge. Removal of strokes or deformities in thinning is a difficult problem. In this paper, we have proposed a nearest opposite contour pixel based thinning strategy used for performing skeletonization of printed and handwritten character images. In this method, we have used shape characteristics of text to get skeleton of nearly same as the true character shape. This approach helps to preserve the local features and true shape of the character images. The proposed algorithm produces one pixel-width thin skeleton. As a by-product of our thinning approach, the skeleton also gets segmented into strokes in vector form. Hence further stroke segmentation is not required. Experiment is done on printed English and Bengali characters and we obtain less spurious branches comparing with other thinning methods without any post-processing.
