Stroke Edge Research Articles

GDB: Gated Convolutions-based Document Binarization

Document binarization is a crucial pre-processing step for various document analysis tasks. However, existing methods fail to accurately capture stroke edges, primarily due to the inherent limitations of vanilla convolutions and the absence of adequate boundary-related supervision during stroke edge extraction. In this paper, we formulate text extraction as the learning of gating values and propose an end-to-end network architecture based on gated convolutions, named GDB, to address the problem of imprecise stroke edge extraction. The gated convolutions enable the selective extraction of stroke feature with different attention. Our proposed framework comprises two stages. Firstly, a coarse sub-network with an extra edge branch is trained to enhance the precision of feature maps by incorporating a priori mask and edge information. Secondly, a refinement sub-network is cascaded to enhance the output of the first stage using gated convolutions based on the sharp edges. To effectively incorporate global information, GDB also integrates a parallelized multi-scale operation that combines local and global features. We conduct comprehensive experiments on ten Document Image Binarization Contest (DIBCO) datasets from 2009 to 2019 and Document Deblurring Datasets. Experimental results show that our proposed methods outperform the state-of-the-art methods across all metrics on average. Extensive ablation studys demonstrate the efficacy of key components. Available codes: https://github.com/Royalvice/GDB.

APST-Flow: A Reversible Network-Based Artistic Painting Style Transfer Method

In recent years, deep generative models have been successfully applied to perform artistic painting style transfer (APST). The difficulties might lie in the loss of reconstructing spatial details and the inefficiency of model convergence caused by the irreversible en-decoder methodology of the existing models. Aiming to this, this paper proposes a Flow-based architecture with both the en-decoder sharing a reversible network configuration. The proposed APST-Flow can efficiently reduce model uncertainty via a compact analysis-synthesis methodology, thereby the generalization performance and the convergence stability are improved. For the generator, a Flow-based network using Wavelet additive coupling (WAC) layers is implemented to extract multi-scale content features. Also, a style checker is used to enhance the global style consistency by minimizing the error between the reconstructed and the input images. To enhance the generated salient details, a loss of adaptive stroke edge is applied in both the global and local model training. The experimental results show that the proposed method improves PSNR by 5%, SSIM by 6.2%, and decreases Style Error by 29.4% over the existing models on the ChipPhi set. The competitive results verify that APST-Flow achieves high-quality generation with less content deviation and enhanced generalization, thereby can be further applied to more APST scenes.

Document Image Binarization using Image Segmentation Technique

Segmentation of text from badly degraded document images is an extremely difficult assignment because of the high inter/Intra variety between the record foundation and the frontal area text of various report pictures. Picture preparing and design acknowledgment algorithms set aside more effort for execution on a solitary center processor. Designs Preparing Unit (GPU) is more mainstream these days because of its speed, programmability, minimal expense and more inbuilt execution centers in it. The primary objective of this exploration work is to make binarization quicker for acknowledgment of a huge number of corrupted report pictures on GPU. In this framework, we give another picture division calculation that every pixel in the picture has its own limit proposed. We are accomplishing equal work on a window of m*n size and separate article pixel of text stroke of that window. The archive text is additionally sectioned by a nearby edge that is assessed dependent on the forces of identified content stroke edge pixels inside a nearby window.

An elimination method for isolated meshes in a road network considering stroke edge feature

The road network is the skeletal element of topographic maps at different scales. In general, urban roads are connected by road segments, thus forming a series of road meshes. Mesh elimination is a key step in evaluating the importance of roads during the road network data management and a prerequisite to the implementation of continuous multiscale spatial representation of road networks. The existing mesh-based method is an advanced road elimination method whereby meshes with the largest density are sequentially selected and road segments with the least importance in each mesh are eliminated. However, the road connectivity and integrity may be destroyed in specific areas by this method because some eliminated road segments could be located in the middle of road strokes. Therefore, this paper proposed an elimination method for isolated meshes in a road network considering stroke edge feature. First, small meshes were identified by using mesh density thresholds, which can be obtained by the sample data statistical algorithm. Thereafter, the small meshes related to the edge segments of road strokes were taken out and defined as stroke edge meshes, and the remaining small meshes were defined as stroke non-edge meshes. Second, by computing the mesh density of all stroke edge meshes, the mesh with the largest density was selected as the starting mesh, and the least important edge segment in the mesh was eliminated. The difference between the existing mesh-based method and the proposed method is that the starting mesh is a stroke edge mesh, not any given small mesh, and the eliminated segment is just only one of edge segments of strokes not chosen from among all segments. Third, mesh elimination was implemented by iteratively processing the stroke edge meshes with the largest mesh density until all of them were eliminated and their mesh density exceeded the threshold. The stroke non-edge meshes were directly preserved. Finally, a 1:10,000 topographic road map of an area in Jiangsu Province of China was used for validation. The experimental results show that for all stroke non-edge meshes and 23% of the stroke edge meshes, compared to the mesh-based method, the road stroke connectivity and integrity of road strokes were better preserved by the proposed method, and the remaining 77% of the elimination results for the stroke edge meshes were the same under the two methods.

SurfaceBrush

Popular Virtual Reality (VR) tools allow users to draw varying-width, ribbonlike 3D brush strokes by moving a hand-held controller in 3D space. Artists frequently use dense collections of such strokes to draw virtual 3D shapes. We propose SurfaceBrush , a surfacing method that converts such VR drawings into user-intended manifold free-form 3D surfaces, providing a novel approach for modeling 3D shapes. The inputs to our method consist of dense collections of artist-drawn stroke ribbons described by the positions and normals of their central polylines, and ribbon widths. These inputs are highly distinct from those handled by existing surfacing frameworks and exhibit different sparsity and error patterns, necessitating a novel surfacing approach. We surface the input stroke drawings by identifying and leveraging local coherence between nearby artist strokes. In particular, we observe that strokes intended to be adjacent on the artist imagined surface often have similar tangent directions along their respective polylines. We leverage this local stroke direction consistency by casting the computation of the user-intended manifold surface as a constrained matching problem on stroke polyline vertices and edges. We first detect and smoothly connect adjacent similarly-directed sequences of stroke edges producing one or more manifold partial surfaces. We then complete the surfacing process by identifying and connecting adjacent similarly directed edges along the borders of these partial surfaces. We confirm the usability of the SurfaceBrush interface and the validity of our drawing analysis via an observational study. We validate our stroke surfacing algorithm by demonstrating an array of manifold surfaces computed by our framework starting from a range of inputs of varying complexity, and by comparing our outputs to reconstructions computed using alternative means.

A Comprehensive Review on Document Image (DIBCO) Database

One of the most significant current discussions in image processing is a document analysis. Now, many types of document database were established in order to address the issue of binarization effectiveness. In this paper, a comprehensive review of the document database was presented. Review based on an image from Document Image Binarization Contest (DIBCO) from 2013 to 2017 which consists of handwritten and printed image. The best algorithm for each year is discussed and analyzed. Based on the results, the technique using background estimation and stroke edges is better performance for the overall database. Besides, the method using the combination of Laplacian operator and canny edge detection also shows the successful result, especially in the printed image. Implications of the review give the direction for future binarization approach developments.

Brain Stroke Segmentation using Fuzzy C-Means Clustering

Image processing technique plays an important role in medical science for envisage various phenomenal structure of human body. Even though it helps more, sometimes it’s very difficult to detect abnormal structures of human body by using simple images. Magnetic Resonance Imaging (MRI) is the one of the most significant technique to analyze human body and helpful for distinguishing and expounding the neural architecture of human brain effectively. This proposed strategy focus on detection and extraction of brain stroke from different patient’s MRI images. In this work some preprocessing techniques like noise removal, filtering and segmentation is used for extract brain stroke partition accurately. The segmentation of brain stroke is implemented by using Fuzzy C-Means (FCM) clustering with two different levels of extraction. Edge detection is used for finding segmented portion of brain stroke edges accurately. Finally the stroke size is calculated for help doctors to make effective decisions about brain stroke. The experimental result proven that the proposed method is successful in detecting and extraction brain stroke efficiently with less time.

Image Binarization for Degraded Document Images

Image binarization is the separation of each pixel values into two collections, black as a foreground and white as a background. Thresholding technique is used for document image binarization. Image binarization plays vital role in segmentation of text from the document images that are badly degraded due to the high inter\intra variations between the foreground text of document images and document background. This paper, proposes technique to address the issues of degraded images using adaptive image contrast. The adaptive image contrast technique is a combination of the local image contrast and the local image gradient. And they are tolerant to variation of text and background. Such variations are caused by number of document degradations. The proposed technique, constructs adaptive contrast map for degraded image .the contrast map is combined with Canny’s edge map, for the identification of text stroke edge pixels. Thresholding technique can be applied as global technique and local technique. Global thresholding is suitable for a document where there is uniform contrast delivery of background and foreground. However global thresholding fails to the applications where difference in contrast, Extensive background noise and difference in brightness exists. in such circumstances categorization of many pixels as a foreground or as a background is not so easy. Local thresholding plays significant role in such cases. Local thresholding technique uses local threshold t; w.r.t .local window to segment the document image .this local threshold t is estimated based on the intensities of detected text stroke edge pixels. The proposed method is simple, robust, and involves minimum parameter tuning. It has been tested on three public datasets that are used in the recent document image binarization contest (DIBCO) 2009 & 2011 and handwritten-DIBCO 2010.

Document Image Binarization Technique for Degraded Document Images

image binarization is a vital pre-processing technique for document image analysis that segments text from badly degraded document images. In this paper, we propose a robust document image binarization technique that is based on the concept of adaptive image contrast. The adaptive image contrast which is formed by combining local image contrast and the local image gradient makes it tolerant to text and background variation caused by different types of document degradations. In the proposed technique the adaptive contrast map is binarized and text stroke edge pixels are detected using Canny's algorithm. The document text is further segmented by a local threshold that is assessed in light of the intensities of detected text stroke edge pixels within a local window. The above mentioned process has been rehashed by combining adaptive image contrast with Sobel's Edge detection technique and Total Variation Edge Detection technique respectively A comparison between these techniques is then made on the basis of Peak-signal to Noise Ratio and Mean Square Error values. These methods have been tested on images suffering from different types of degradations .It has been found out that adaptive image contrast used with Canny's edge detection technique gives the best results.

A Novel Document Image Binarization For Optical Character Recognition

This paper presents a technique for document image binarization that segments the foreground text accurately from poorly degraded document images. The proposed technique is based on the Segmentation of text from poorly degraded document images and it is a very demanding job due to the high variation between the background and the foreground of the document. This paper proposes a novel document image binarization technique that segments the texts by using adaptive image contrast. It is a combination of the local image contrast and the local image gradient that is efficient to overcome variations in text and background caused by different types degradation effects. In the proposed technique, first an adaptive contrast map is constructed for a degraded input document image. The contrast map is then binarized by global thresholding and pooled with Canny’s edge map detection to identify the text stroke edge pixels. By applying Segmentation the text is further segmented by a local thresholding method that. The proposed method is simple, strong, and requires minimum parameter tuning.

Uygur handwriting identification based on edge stroke feature structure

According to the features of Uygur letters,the author proposed a method of identifying the stroke edge quantification model of handwriting.On the basis of extracting edge image,a text-independent,direction and length-dependent feature structure vector model was set up on the four-class angles tendency of Uygur letter handwriting edge by using the basic stroke concept of "horizontal,vertical,left-descending and right-descending strokes".The author counted all feature structures of local windows to obtain the probability density feature vector of the edge stroke,used weighted and unweighted distance formulas to get the handwriting feature vector distance between identification and reference samples and judged candidate writers by sorting vector distance.With a relative strong practicability,this method can describe local feature and style of Uygur letter handwriting commendably and achieve better identification results.

Robust Document Image Binarization Technique for Degraded Document Images

Segmentation of text from badly degraded document images is a very challenging task due to the high inter/intra-variation between the document background and the foreground text of different document images. In this paper, we propose a novel document image binarization technique that addresses these issues by using adaptive image contrast. The adaptive image contrast is a combination of the local image contrast and the local image gradient that is tolerant to text and background variation caused by different types of document degradations. In the proposed technique, an adaptive contrast map is first constructed for an input degraded document image. The contrast map is then binarized and combined with Canny's edge map to identify the text stroke edge pixels. The document text is further segmented by a local threshold that is estimated based on the intensities of detected text stroke edge pixels within a local window. The proposed method is simple, robust, and involves minimum parameter tuning. It has been tested on three public datasets that are used in the recent document image binarization contest (DIBCO) 2009 & 2011 and handwritten-DIBCO 2010 and achieves accuracies of 93.5%, 87.8%, and 92.03%, respectively, that are significantly higher than or close to that of the best-performing methods reported in the three contests. Experiments on the Bickley diary dataset that consists of several challenging bad quality document images also show the superior performance of our proposed method, compared with other techniques.

Document image binarization using background estimation and stroke edges

Document images often suffer from different types of degradation that renders the document image binarization a challenging task. This paper presents a document image binarization technique that segments the text from badly degraded document images accurately. The proposed technique is based on the observations that the text documents usually have a document background of the uniform color and texture and the document text within it has a different intensity level compared with the surrounding document background. Given a document image, the proposed technique first estimates a document background surface through an iterative polynomial smoothing procedure. Different types of document degradation are then compensated by using the estimated document background surface. The text stroke edge is further detected from the compensated document image by using L1-norm image gradient. Finally, the document text is segmented by a local threshold that is estimated based on the detected text stroke edges. The proposed technique was submitted to the recent document image binarization contest (DIBCO) held under the framework of ICDAR 2009 and has achieved the top performance among 43 algorithms that are submitted from 35 international research groups.