Candidate Text Regions Research Articles

Advertising Image Saliency Prediction Method Based on Score Level Fusion

At present, visual saliency prediction algorithms have been developed more and more mature, but most of the current saliency prediction algorithms are aimed at natural images. Due to the inconsistency of elements and features between natural images and advertising images, the existing saliency prediction algorithms show poor robustness and low inference speed to advertising images, which severely limits its commercial application in advertising design and evaluation. In view of this, a saliency prediction algorithm for advertisement images is proposed in this paper. In the feature extraction stage, two text candidate regions based on intensity feature and improved MESR algorithm are first obtained and further integrated to produce a two-dimensional text confidence score. Meanwhile, a saliency confidence score is also obtained by an improved natural image saliency prediction network. Then, the score level fusion strategy was adopted to fuse the two confidence scores to get the final saliency prediction map. The experimental results show that the proposed model has good accuracy and robustness in advertising images, as well as the most remarkable inference speed, which can meet the demand for real-time performance of advertising image saliency prediction, leading to great practical and commercial value.

An Effective Method for Detection and Recognition of Uyghur Texts in Images with Backgrounds

Uyghur text detection and recognition in images with simple backgrounds is still a challenging task for Uyghur image content analysis. In this paper, we propose a new effective Uyghur text detection method based on channel-enhanced MSERs and the CNN classification model. In order to extract more complete text components, a new text candidate region extraction algorithm is put forward, which is based on the channel-enhanced MSERs according to the characteristics of Uyghur text. In order to effectively prune the non-text regions, we design a CNN classification network according to the LeNet-5, which gains the description characteristics automatically and avoids the tedious and low efficiency artificial characteristic extraction work. For Uyghur text recognition in images, we improved the traditional CRNN network, and to verify its effectiveness, the networks trained on a synthetic dataset and evaluated on the text recognition datasets. The experimental results indicated that the Uyghur text detection method in this paper is robust and applicable, and the recognition result by improvedCRNN was better than the original CRNN network.

Urdu text in natural scene images: a new dataset and preliminary text detection.

Text detection in natural scene images for content analysis is an interesting task. The research community has seen some great developments for English/Mandarin text detection. However, Urdu text extraction in natural scene images is a task not well addressed. In this work, firstly, a new dataset is introduced for Urdu text in natural scene images. The dataset comprises of 500 standalone images acquired from real scenes. Secondly, the channel enhanced Maximally Stable Extremal Region (MSER) method is applied to extract Urdu text regions as candidates in an image. Two-stage filtering mechanism is applied to eliminate non-candidate regions. In the first stage, text and noise are classified based on their geometric properties. In the second stage, a support vector machine classifier is trained to discard non-text candidate regions. After this, text candidate regions are linked using centroid-based vertical and horizontal distances. Text lines are further analyzed by a different classifier based on HOG features to remove non-text regions. Extensive experimentation is performed on the locally developed dataset to evaluate the performance. The experimental results show good performance on test set images. The dataset will be made available for research use. To the best of our knowledge, the work is the first of its kind for the Urdu language and would provide a good dataset for free research use and serve as a baseline performance on the task of Urdu text extraction.

Scene Text Detection with Quadtree-based Candidate Text Regions and Convolutional Neural Network

Scene Text Detection with Quadtree-based Candidate Text Regions and Convolutional Neural Network

An Enhanced MSER Pruning Algorithm for Detection and Localization of Bangla Texts from Scene Images

Text detection and localization have great importance for content based image analysis and text based image indexing. The efficiency of text recognition depends on the efficiency of text localization. So, the main goal of the proposed method is to detect and localize text regions with high accuracy. To achieve this goal, a new and efficient method has been introduced for localization of Bangla text from scene images. In order to improve precision and recall as well as f-measure, Maximally Stable Extremal Region (MSER) based method along with double filtering techniques have been used. As MSER algorithm generates many false positives, we have introduced double filtering method for removing these false positives to increase the f-measure to a great extent. Our proposed method works at three basic levels. Firstly, MSER regions are generated from the input color image by converting it into gray scale image. Secondly, some heuristic features are used to filter out most of the false positives or non-text regions. Lastly, Stroke Width Transform (SWT) based filtering method is used to filter out remaining non-text regions. Remaining components are then grouped into candidate text regions marked by bounding box over each region. As there is no benchmark database for Bangla text, the proposed method is implemented on our own prepared database consisting of 200 scene images of Bangla texts and has got prominent performance. To evaluate the performance of our proposed approach, we have also tested the proposed method on International Conference on Document Analysis and Recognition( ICDAR) 2013 benchmark database and have got a better result than the related existing methods.

Multi-Lingual Scene Text Detection Using One-Class Classifier

The main purpose of scene text recognition is to detect texts in a given image. The problem of text detection and recognition in such images has gained great attention in recent years due to rising demand of several applications like visual based applications, multimedia and content-based retrieval. Due to low accuracies of existing scene text detection methods, an improved pipeline is developed for text localizing task. First, candidate text regions are generated using Maximally Stable Extremal Region and Stroke Width Transform methods that capture true positives along with many false positives. A One Class Classifier is trained to label the candidate regions obtained, as text or non-text, which in this case is suitable as non-text class cannot be adequately represented to train a binary classifier. The one class classifier is trained with some popular feature descriptors like Histogram of Oriented Gradients, Grey Level Co-Occurrence Matrix, Discrete Cosine Transform and Gabor filter. Experimental results show high recall for text containing regions and reducing false positives.

Scene Text Detection Using Superpixel-Based Stroke Feature Transform and Deep Learning Based Region Classification

Scene text detection is a crucial step in end-to-end scene text recognition, a greatly challenging problem in computer vision. This paper proposes a novel scene text detection method that involves superpixel-based stroke feature transform (SSFT) and deep learning based region classification (DLRC). The SSFT is developed for candidate character region (CCR) extraction, which consists in partitioning an input image into several regions via superpixel-based clustering, removing most regions based on predefined criteria satisfied by the characters, and refining the remaining regions to obtain CCRs by computing a stroke width map. The character regions are identified from the CCRs using DLRC, in which several hand-crafted low-level features, i.e., color, texture, and geometric features, and some deep convolution neural network (CNN) based high-level features are first extracted from the regions, and then these features are fused by using two fully connected networks (FCNs) for region classification. In the DLRC step, the deep feature extraction CNN and the feature fusion FCNs are jointly trained. Next, the extracted character regions are merged to form candidate text regions, from which the final scene texts are detected. The proposed method is evaluated on three publicly available datasets: ICDAR2011, ICDAR2013, and street view text. It achieves F -measures of 0.876, 0.885, and 0.631, respectively, which demonstrate the effectiveness of the proposed scene text detection method.

Study on the extraction of the text region from natural scene images by an analysis of the edge-oriented pattern

Background/Objectives: In this paper, we propose a hybrid scene-detection method using an edge and textural analysis in natural scene images, and finally, we detect the text regions by removing the non-text regions through a pattern analysis of each region.Methods/Statistical analysis: The proposed algorithm is divided into the pre-processing stage and the extraction processing stage to perform the text detection. The lost texts that are improved through a histogram equalization for the minimization of the loss of the text parts that is due to light exposure are detected before the edge detection. After that, the edge is detected using the Canny operator. The detected edge is obtained in the step of applying the SWT algorithm to detect the text candidate regions. The extraction processing step is the step of removing the noise region that is detected by the pixel analysis of the SWT algorithm, and it analyzes the pattern of the text regions and then removes the non-text regions to finally detect the text regions. For the quantitative comparison of the proposed algorithm, our results are compared with the ground-truth image using the precision, recall, and F-measure.Findings: One of the existing text-detection algorithms, the edge-based method, is problematic, as, in addition to the text, the complex backgrounds and textures are detected as the edges in natural scene images. The connected component-based method is also problematic, as the non-text region is included in the text region in the process of finding the connection component.Improvements/Applications: The proposed method shows an effective text-detection result regardless of the light exposure in natural scene images compared with the conventional text-detection algorithm.

A Heuristic Approach to Detect and Localize Text on Arabic News Video

Automatic text detection in video sequences remains a challenging problem due to the variety of sizes, colors and the presence of complex background. In this paper, we attempt to solve this problem by proposing a robust detection–validation schema for text localization in Arabic news video. Candidate text regions are first detected by using a hybrid method which combines MSER detector and edge information. Then, these regions are grouped using morphological operators. Finally, a verification process is applied to remove noisy non-text regions including specific features for Arabic text. Performance and efficacy of the proposed text detection approach have been tested By using Arabic-Text-in-Video database (AcTiV-DB).

A Novel Approach for Video Text Detection and Recognition Based on a Corner Response Feature Map and Transferred Deep Convolutional Neural Network

The text presented in videos contains important information for content analysis, indexing, and retrieval of videos. The key technique for extracting this information is to find, verify, and recognize video text in various languages and fonts against complex backgrounds. In this paper, we propose a novel method that combines a corner response feature map and transferred deep convolutional neural networks for detecting and recognizing video text. First, we use a corner response feature map to detect candidate text regions with a high recall. Next, we partition the candidate text regions into candidate text lines by projection analysis using two alternative methods. We then construct classification networks transferred from VGG16, ResNet50, and InceptionV3 to eliminate false positives. Finally, we develop a novel fuzzy c-means clustering-based separation algorithm to obtain a clean text layer from complex backgrounds so that the text is correctly recognized by commercial optical character recognition software. The proposed method is robust and has good performance on video text detection and recognition, which was evaluated on three publicly available test data sets and on the high-resolution test data set we constructed.

An Improved Text Localization Method for Natural Scene Images

In order to extract text information effectively from natural scene image with complex background, multi-orientation perspective and multilingual languages, we present a new method based on the improved Stroke Feature Transform (SWT). Firstly, The Maximally Stable Extremal Region (MSER) method is used to detect text candidate regions. Secondly, the SWT algorithm is used in the candidate regions, which can improve the edge detection compared with tradition SWT method. Finally, the Frequency-tuned (FT) visual saliency is introduced to remove non-text candidate regions. The experiment results show that, the method can achieve good robustness for complex background with multi-orientation perspective, various characters and font sizes.

Deteksi Teks Secara Otomatis Pada Natural Image Berbasis Superpixel Menggunakan Maximally Stable Extremal Regions dan Stroke Width Transform

Text detection in natural image is something to do before performing character recognition. The process of text detection plays an important role in the acquisition of information in an image. This research aims to detect text automatically in natural image based on superpixels with Maximally Stable Extremal Regions (MSER) and Stroke Width Transform (SWT). The superpixel method used is Simple Linear Iterative Clustering (SLIC). The SLIC method is used for segmenting text images into superpixel spaces. Image segmentation to superpixel aims to group pixels into homogeneous regions that capture redundant images. SLIC is a technique that effectively divides images into homogeneous regions (superpixels). Furthermore MSER is used as a feature to locate the text candidate region in a segmented image with superpixel. Then edge detection is done to validate the text area that has been found. Next, the SWT method is used to distinguish both text and non-text image regions. The dataset used is ICDAR 2003. Based on test result, MSER with superpixel is able to detect region of text in natural image. SWT is also able to recover the region which is the candidate of the text in natural image.

Edge color transform: a new operator for natural scene text localization

Text extraction is one of the most intuitive and natural ways in understanding of natural scene images. Localization is the first and one of the most important steps in this problem. In this paper, we propose a novel method for localizing text in natural scene images. We introduce a new operator so called Edge Color Transform (ECT) to solve this problem. After extracting the edge map of the input image, this operator follows the gradient direction (and the opposite direction) to assign the nearest color to each of the edge pixels. In the next step, each channel (Red, Green and Blue) of these color pixels are normalized to make them more robust to intensity changes. In the following step, these colored pixels are grouped together using a modified version of region growing algorithm. In this customized version of the algorithm, pixels of a region do not have to be spatially connected and their connectivity is decided based on a predefined oval-shaped neighborhood. When the regions are forged, each of them form a text candidate region. Subsequently, all of these candidates are applied to a classifier to extract text regions. Experimental results show that the proposed method performs well on both Farsi and well-known ICDAR 2013 data sets.

A Robust Approach for Scene Text Localization Using Rule-Based Confidence Map and Grouping

This study presents a robust algorithm to localize both Farsi/Arabic and Latin scene texts with different sizes, fonts and orientations even the low luminance contrast and poor quality ones. First, a new region detector is proposed to extract the candidate text regions. It is an integration of the weighted median filtering, contrast preserving decolorization and MSER techniques. It is robust to the low luminance contrast and poor quality scene texts. Afterwards, a novel method based on the fuzzy inference systems (FIS) is proposed to build a confidence map. This map indicates the likelihood of being text for the extracted candidate regions. Therefore it is exploited to filter the nontext candidates. Finally, a new fuzzy-based approach is proposed to create the single arbitrarily oriented text lines. It is based on the clustering, FIS, minimum area rectangle as well as radon transform techniques. It could also retrieve some of the discarded isolated characters or subwords in the filtering stage. To validate the proposed algorithm, we created a collection of natural images containing both Farsi/Arabic and Latin texts. Compared with the state-of-the-art methods, the proposed method achieves the best performance on our and Epshtein datasets and competitive performances on the ICDAR dataset.

Scene Text Detection and Segmentation based on Cascaded Convolution Neural Networks.

Scene text detection and segmentation are two important and challenging research problems in the field of computer vision. This paper proposes a novel method for scene text detection and segmentation based on cascaded convolution neural networks (CNNs). In this method, a CNN based text-aware candidate text region (CTR) extraction model (named detection network, DNet) is designed and trained using both the edges and the whole regions of text, with which coarse CTRs are detected. A CNN based CTR refinement model (named segmentation network, SNet) is then constructed to precisely segment the coarse CTRs into text to get the refined CTRs. With DNet and SNet, much fewer CTRs are extracted than with traditional approaches while more true text regions are kept. The refined CTRs are finally classified using a CNN based CTR classification model (named classification network, CNet) to get the final text regions. All of these CNN based models are modified from VGGNet-16. Extensive experiments on three benchmark datasets demonstrate that the proposed method achieves state-of-the-art performance and greatly outperforms other scene text detection and segmentation approaches.

Multilingual Artificial Text Extraction and Script Identification from Video Images

This work presents a system for extraction and script identification of multilingual artificial text appearing in video images. As opposed to most of the existing text extraction systems which target textual occurrences in a particular script or language, we have proposed a generic multilingual text extraction system that relies on a combination of unsupervised and supervised techniques. The unsupervised approach is based on application of image analysis techniques which exploit the contrast, alignment and geometrical properties of text and identify candidate text regions in an image. Potential text regions are then validated by an Artificial Neural Network (ANN) using a set of features computed from Gray Level Co-occurrence Matrices (GLCM). The script of the extracted text is finally identified using texture features based on Local Binary Patterns (LBP). The proposed system was evaluated on video images containing textual occurrences in five different languages including English, Urdu, Hindi, Chinese and Arabic. The promising results of the experimental evaluations validate the effectiveness of the proposed system for text extraction and script identification.

Localizing scene texts by fuzzy inference systems and low rank matrix recovery model

In this paper a framework is proposed to localize both Farsi/Arabic and Latin scene texts with different sizes, fonts and orientations. First, candidate text regions are extracted via an MSER detector enhanced by weighted median filtering to adopt the low resolution texts. At the same time based on fuzzy inference system (FIS), the input image is classified into images with a focused text content and incidental scene text images which the image does not focus on the text content. For the focused scene text images the non-text candidates are filtered via an FIS. On the other hand, for the incidental scene text images apart from the FIS, an extra filtering algorithm based on low rank matrix recovery is proposed. Finally, a new approach based on the clustering, minimum area rectangle and radon transform techniques is proposed to create the single arbitrarily oriented text lines from the remaining text regions. To evaluate the proposed algorithm, we created a collection of natural images containing both Farsi/Arabic and Latin texts. Compared with the state-of-the-art methods, the proposed method achieves the best performance on our and Epshtein datasets and competitive performances on the ICDAR dataset.

Multi-Spectral Fusion Based Approach for Arbitrarily Oriented Scene Text Detection in Video Images.

Scene text detection from video as well as natural scene images is challenging due to the variations in background, contrast, text type, font type, font size, and so on. Besides, arbitrary orientations of texts with multi-scripts add more complexity to the problem. The proposed approach introduces a new idea of convolving Laplacian with wavelet sub-bands at different levels in the frequency domain for enhancing low resolution text pixels. Then, the results obtained from different sub-bands (spectral) are fused for detecting candidate text pixels. We explore maxima stable extreme regions along with stroke width transform for detecting candidate text regions. Text alignment is done based on the distance between the nearest neighbor clusters of candidate text regions. In addition, the approach presents a new symmetry driven nearest neighbor for restoring full text lines. We conduct experiments on our collected video data as well as several benchmark data sets, such as ICDAR 2011, ICDAR 2013, and MSRA-TD500 to evaluate the proposed method. The proposed approach is compared with the state-of-the-art methods to show its superiority to the existing methods.

Detecting natural scenes text via auto image partition, two-stage grouping and two-layer classification

Text detection in natural scene images is important and challenging work for image analysis. In this paper, we present a robust system to detect natural scene text according to text region appearances. The framework includes three parts: auto image partition, two-stage grouping and two-layer classification. The first part partitions images into unconstrained sub-images through statistical distribution of sampling points. The designed two-stage grouping method performs grouping in each sub-image in first stage and connects different partitioned image regions in second stage to group connected components (CCs) to text regions. Then a two-layer classification mechanism is designed for classifying candidate text regions. The first layer is to compute the similarity score of region blocks and the second layer is a SVM classifier using HOG features. We add a normalization step to rectify perspective distortion before candidate text region classification which improves the accuracy and robustness of the final output result. The proposed system is evaluated on four types datasets including two ICDAR Robust Reading Competition datasets, a born-digital image dataset, a video image dataset and a perspective distortion image dataset. The experimental results demonstrate that our proposed framework outperforms state-of-the-art localization algorithms and is robust in dealing with multiple background outliers.

Caption Region Detection in Video Images by Improved Corner Detector

Caption region detection is significant for text segmentation and recognition in video images. In this paper, caption region detection includes the following three steps: text corner detection, candidate text region generation, non-text region elimination. Firstly, an improved Harris corner detection algorithm based on second difference, called sHarris, is proposed to detect text corners. sHarris detector is effective to detect text corners existing in video image. Secondly, morphologic operations are utilized to generate candidate text regions after text corner points are detected. Finally, non-text regions are eliminated from candidate text regions by some validation rules. In the end of this paper, experiments are used to analyze and evaluate the detection results of caption regions, which indicate the effectiveness of the proposed method.