Soharab Hossain Shaikh

A comprehensive survey of human action recognition with spatio-temporal interest point (STIP) detector

Over the past two decades, human action recognition from video has been an important area of research in computer vision. Its applications include surveillance systems, human–computer interactions and various real-world applications where one of the actor is a human being. A number of review works have been done by several researchers in the context of human action recognition. However, it is found that there is a gap in literature when it comes to methodologies of STIP-based detector for human action recognition. This paper presents a comprehensive review on STIP-based methods for human action recognition. STIP-based detectors are robust in detecting interest points from video in spatio-temporal domain. This paper also summarizes related public datasets useful for comparing performances of various techniques.

A new image binarization method using iterative partitioning

This paper proposes a new method for image binarization that uses an iterative partitioning approach. The proposed method has been tested towards binarization of both document and graphic images. The quantitative comparisons with other standard methods reveal that the proposed approach outperforms existing widely used binarization techniques in terms of accuracy of binarization. The experimental results further establish the superiority of the proposed method, especially for degraded documents and graphic images. The proposed algorithm is suitable for a multi-core processing environment as it can be split into multiple parallel units of executions after the initial partitioning.

A Comprehensive Survey on Image Binarization Techniques

A detailed survey about the principles of image binarization techniques is introduced in this chapter. A comprehensive review is given. A number of classical methodologies together with the recent works are considered for comparison and study of the concept of binarization for both document and graphic images.

Image binarization using iterative partitioning: A global thresholding approach

This paper proposes a new method for image binarization. This is a modified and improved version of the iterative partition based algorithm proposed in [1]. The proposed method has been compared with other five representative binarization methods including the algorithm proposed in [1]. The USC-SIPI image database has been used for experimental verification purposes. The results of implementation of the algorithms unearth the superiority of the proposed method compared to the other five methods in terms of two quantitative measures, namely, misclassification error and the relative foreground area error.

Performance Evaluation of Multiple Image Binarization Algorithms Using Multiple Metrics on Standard Image Databases

The area of image binarization has matured to a significant extent in last few years. There has been multiple, well-defined metrics for quantitative performance estimation of the existing techniques for binarization. However, it stills remains a problem to benchmark one binarization technique with another as different metrics are used to establish the comparative edges of different binarization approaches. In this paper, an experimental work is reported that uses three different metrics for quantitative performance evaluation of seven binarization techniques applied on four different types of images: Arial, Texture, Degraded text and MRI. Based on visually and experimentally the most appropriate methods for binarization of images have been identified for each of the four classes under consideration. We have used standard image databases along with the archived reference images, as available, for experimental purpose.

Automatic Detection and Classification of Solitary Pulmonary Nodules from Lung CT Images

Cancer is one of the fatal diseases, posing threat to human life. Among different types of cancer, lung cancer can be considered as one of the most most deadly one. Lung nodules are small white spots that appear in lung parenchyma. Lung nodules are primarily of two types, solitary pulmonary nodules and juxtrapleural nodules. Solitary pulmonary nodules are round in shape, whereas juxtapleural nodules have a worm like shape, which are generally introduced through metathesis from the other cancerous organ of the human body. All lung nodules are not cancerous. Each of them has certain geometric features, upon which the nodules can be classified into cancerous and non-cancerous. Chest radiographs and computed tomography (CT) scan images are important for the purpose of diagnosis of the lung cancer. Manual detection of cancerous lung nodule is a very challenging task as it is time consuming and prone to human error. An automated detection is therefore needed for faster detection of malignant and benign lung nodules based on the shape features of the nodules. This automated detection will also help to reduce the cost of diagnosis by selectively choosing only the malignant nodule for biopsy tissue culture discarding the benign lung nodule. In this paper, we have proposed a novel method that detects and categorizes solitary pulmonary nodules responsible for lung cancer from lung CT images. Our method reduces variability in detection by automatic segmentation and classification of nodules. The experimental results are promising in respect to classification of lung nodules as malignant or benign.

A Low Cost Moving Object Detection Method Using Boundary Tracking

Moving object detection techniques have been studied extensively for such purposes as video content analysis as well as for remote surveillance. Video surveillance systems rely on the ability to detect moving objects in the video stream which is a relevant information extraction step in a wide range of computer vision applications. There are many ways to track the moving object. Most of them use the frame differences to analyze the moving object and obtain object boundary. This may be quite resource hungry in the sense that such approaches require a large space and a lot of time for processing. This paper proposes a new method for moving object detection from video sequences by performing frame-boundary tracking and active-window processing leading to improved performance with respect to computation time and amount of memory requirements. A stationary camera with static background is assumed.

Moving Object Detection Approaches, Challenges and Object Tracking

There are various approaches to moving object detection from video; e.g. background subtraction, temporal differencing, statistical approaches, optical flow etc. This chapter summarizes these methodologies. Different challenging conditions that pose problems in moving object detection are also identified. Object tracking, a task closely related to moving object detection is also discussed in brief in this chapter.

Modified Majority Voting Algorithm towards Creating Reference Image for Binarization

The quantitative evaluation of different binarization techniques to measure their comparative performance is indeed an important aspect towards avoiding subjective evaluation. However, in majority of the papers found in the literature, creating a reference image is based on manual processing. These are often highly subjective and prone to human error. No single binarization technique so far has been found to produce consistently good results for all types of textual and graphic images. Thus creating a reference image indeed remains an unsolved problem. As found in the majority voting approach, a strong bias, due to poor computation of threshold by one or two methods for a particular image, has often had an adverse effect in computing the threshold for the reference image. The improvement proposed in this paper helps eliminate this bias to a great extent. Experimental verification using images from a standard database illustrates the effectiveness of the proposed method.

Applications of Binarization

This chapter shows the applications of binarization in image processing and analysis. A particular consideration is given toward some selected biometric applications. Use of binarization in medical image processing is also stated.