Lihong Zheng

An algorithm for accuracy enhancement of license plate recognition

This paper presents an algorithm for extraction (detection) and recognition of license plates in traffic video datasets. For license plate detection, we introduce a method that applies both global edge features and local Haar-like features to construct a cascaded classifier consisting of 6 layers with 160 features. The characters on a license plate image are extracted by a method based on an improved blob detection algorithm for removal of unwanted areas. For license plate recognition (i.e., character recognition), an open source OCR is modified and used. Our proposed system is robust under poor illumination conditions and for moving vehicles. Our overall system is efficient and can be applied in real-time applications. Experimental results are demonstrated using a traffic video.

Accuracy Enhancement for License Plate Recognition

Automatic License Plate Recognition is useful for real time traffic management and surveillance. License plate recognition usually contains two steps, namely license plate detection/localization and character recognition. Recognizing characters in a license plate is a very difficult task due to poor illumination conditions and rapid motion of vehicles. When using an OCR for character recognition, it is crucial to correctly remove the license plate boundaries after the step for license plate detection. No matter which OCRs are used, the recognition accuracy will be significantly reduced if the boundaries are not properly removed. This paper presents an efficient algorithm for non character area removal. The algorithm is based on the license plates detected using an AdaBoost algorithm. Then it follows the steps of character height estimation, character width estimation, segmentation and block identification. The algorithm is efficient and can be applied in real-time applications. The experiments are performed using OCR software for character recognition. It is shown that much higher recognition accuracy is obtained by gradually removing the license plate boundaries.

Segmentation of characters on car license plates

License plate recognition usually contains three steps, namely license plate detection/localization, character segmentation and character recognition. When reading characters on a license plate one by one after license plate detection step, it is crucial to accurately segment the characters. The segmentation step may be affected by many factors such as license plate boundaries (frames). The recognition accuracy will be significantly reduced if the characters are not properly segmented. This paper presents an efficient algorithm for character segmentation on a license plate. The algorithm follows the step that detects the license plates using an AdaBoost algorithm. It is based on an efficient and accurate skew and slant correction of license plates, and works together with boundary (frame) removal of license plates. The algorithm is efficient and can be applied in real-time applications. The experiments are performed to show the accuracy of segmentation.

Supervised Latent Linear Gaussian Process Latent Variable Model for Dimensionality Reduction

The Gaussian process (GP) latent variable model (GPLVM) has the capability of learning low-dimensional manifold from highly nonlinear data of high dimensionality. As an unsupervised dimensionality reduction (DR) algorithm, the GPLVM has been successfully applied in many areas. However, in its current setting, GPLVM is unable to use label information, which is available for many tasks; therefore, researchers proposed many kinds of extensions to the GPLVM in order to utilize extra information, among which the supervised GPLVM (SGPLVM) has shown better performance compared with other SGPLVM extensions. However, the SGPLVM suffers in its high computational complexity. Bearing in mind the issues of the complexity and the need of incorporating additionally available information, in this paper, we propose a novel SGPLVM, called supervised latent linear GPLVM (SLLGPLVM). Our approach is motivated by both SGPLVM and supervised probabilistic principal component analysis (SPPCA). The proposed SLLGPLVM can be viewed as an appropriate compromise between the SGPLVM and the SPPCA. Furthermore, it is also appropriate to interpret the SLLGPLVM as a semiparametric regression model for supervised DR by making use of the GP to model the unknown smooth link function. Complexity analysis and experiments show that the developed SLLGPLVM outperforms the SGPLVM not only in the computational complexity but also in its accuracy. We also compared the SLLGPLVM with two classical supervised classifiers, i.e., a GP classifier and a support vector machine, to illustrate the advantages of the proposed model.

A Survey on Human Action Recognition Using Depth Sensors

The recent advent of depth sensors opens up new opportunities to advance human action recognition by providing depth information. Many different approaches have been proposed for human action recognition using depth sensors. The main purpose of this paper is to provide a comprehensive study and an updated review on human action recognition using depth sensors. We give an overview of recent works in this field from the viewpoints of data modalities, feature extraction and classification. In terms of data modalities from depth sensors, recent approaches can be roughly categorized into depth map based and skeleton based approaches. Since depth maps encode 3D shape and appearance information, approaches based on depth maps are suitable for short simple actions and can achieve high performance. In contrast, due to the discriminative power and more concise form of skeletal joints, skeleton based approaches can model more complex actions, even in real time. This paper further provides a summary of the results obtained in the last couple of years on the public datasets. Moreover, we discuss limitations of the state of the art and outline promising directions of research in this area. The review assists in guiding both researchers and practitioners in the selection and development of approaches for human action recognition using depth sensors.

3D Motion Trail Model Based Pyramid Histograms of Oriented Gradient for Action Recognition

Human action recognition based on the depth maps is an important yet challenging task. In this paper, a new framework based on the 3D motion trail model (3DMTM) and Pyramid Histograms of Oriented Gradient (PHOG) is proposed to recognize human actions from sequences of depth maps. Specifically, a discriminative descriptor called 3DMTM-PHOG is proposed for depth-based human action recognition. The 3DMTM is generated through the entire depth video sequence to encode additional motion information from three projected orthogonal planes. By adding pyramid representation, Histograms of Oriented Gradient (HOG) descriptor is extended to PHOG which can well characterize local shapes at different spatial grid sizes for action recognition. PHOG is then computed from the 3DMTM as the 3DMTM-PHOG descriptor for the representation of an action. The proposed approach based on 3DMTM-PHOG descriptor is evaluated on MSR Action3D dataset captured by depth cameras. Experimental results show that the proposed approach outperforms the state-of-the-art methods and demonstrate the effectiveness and robustness of the proposed 3DMTM-PHOG descriptor.

Multi-modal Gesture Recognition Using Skeletal Joints and Motion Trail Model

This paper proposes a novel approach to multi-modal gesture recognition by using skeletal joints and motion trail model. The approach includes two modules, i.e. spotting and recognition. In the spotting module, a continuous gesture sequence is segmented into individual gesture intervals based on hand joint positions within a sliding window. In the recognition module, three models are combined to classify each gesture interval into one gesture category. For skeletal model, Hidden Markov Models (HMM) and Support Vector Machines (SVM) are adopted for classifying skeleton features. For depth maps and user masks, we employ 2D Motion Trail Model (2DMTM) for gesture representation to capture motion region information. SVM is then used to classify Pyramid Histograms of Oriented Gradient (PHOG) features from 2DMTM. These three models are complementary to each other. Finally, a fusion scheme incorporates the probability weights of each classifier for gesture recognition. The proposed approach is evaluated on the 2014 ChaLearn Multi-modal Gesture Recognition Challenge dataset. Experimental results demonstrate that the proposed approach using combined models outperforms single-modal approaches, and the recognition module can perform effectively on user-independent gesture recognition.

D 2 MAC: Dynamic delayed Medium Access Control (MAC) protocol with fuzzy technique for Wireless Body Area Networks

Wireless Body Area Networks (WBAN) have emerged as an extension to conventional wireless sensor networks in recent years to comply with the needs in providing timely and effective response in healthcare as one of the many target applications such networks have. The traffic of a WBAN is diverse due to different monitoring tasks carried on by sensor nodes. It brings difficulty in how to efficiently organize the access to the medium for the dynamic and various generated traffic. This paper analyses the traffic diversity problem in WBAN for healthcare applications and proposes a dynamic delayed Medium Access Control (MAC) algorithm. A fuzzy logic system is used to incorporate both application and protocol related parameters of the real time traffic to make the backoff time produced in IEEE 802.15.4 MAC protocol traffic adaptive. The simulation results demonstrate a significant reliability in packet transmissions and decrease in the latency with no change in energy consumption level.

Three Dimensional Motion Trail Model for Gesture Recognition

In this paper an effective method is presented to recognize human gestures from sequences of depth images. Specifically, we propose a three dimensional motion trail model (3D-MTM) to explicitly represent the dynamics and statics of gestures in 3D space. In 2D space, the motion trail model (2D-MTM) consists of both motion information and static posture information over the gesture sequence along the xoy-plane. Considering gestures are performed in 3D space, depth images are projected onto two other planes to encode additional gesture information. The 2D-MTM is then extensively combined with complementary motion information from additional two planes to generate the 3D-MTM. Furthermore, the Histogram of Oriented Gradient (HOG) feature vector is extracted from the proposed 3D-MTM as the representation of a gesture sequence. The experiment results show that the proposed method achieves better results on two publicly available datasets namely MSR Action3D dataset and ChaLearn gesture dataset.

Dynamic backoff scheduling of low data rate applications in wireless body area networks

AbstractnReliability stands as the first important factor when dealing with medical data within the context of a wireless body area network (WBAN). The sensor nodes on body send their data to the coordinator based on a beacon-enabled access mode defined in IEEE 802.15.4 medium access control (MAC) specifications. This paper studies an effective backoff resolution into the carrier sense multiple access with collision avoidance (CSMA/CA) procedure of the beacon-enabled access mode of IEEE 802.15.4. Whilst the standard introduces the backoff as a resolution to less probability of identical backoffs, it does not address the efficiency of the generated backoff time in non-identical backoff situations. This phenomenon degrades the reliability of the received data at the coordinator device. A dynamic and reliable MAC algorithm for WBAN is presented in which length of the backoff period for each node gets decided based on its past successful trials in accessing the channel and also its data rate. This encourages a fair access to the medium among all the sensor nodes as moderate backoff values are assigned to each node. The primary contributions in this paper are less delay endured and higher data reliability while making no changes to the level of energy efficiency.n