Search

In error‐prone wireless environments, it is difficult to realize video coding systems that are robust to various types of data loss. In this paper, a novel motion‐vector refinement approach is presented for video error concealment. A traditional boundary‐matching approach is exploited to reduce blocky effects along the block boundary. More specifically, a downhill simplex approach is combined with a boundary‐matching approach to fine‐tune the motion vectors, reducing the blocky effects along the prediction unit block boundary, and minimizing the computational cost. Extensive simulations are performed, and the results obtained verify the robustness and effectiveness of the proposed approach.

The general method of machine learning has encountered disadvantages in terms of the significant amount of time and effort required for feature extraction and engineering in natural language processing. However, in recent years, these disadvantages have been solved using deep learning. In this paper, we propose a mention pair (MP) model using deep learning, and a system that combines both rule-based and deep learning-based systems using a guided MP as a coreference resolution, which is an information extraction technique. Our experiment results confirm that the proposed deep-learning based coreference resolution system achieves a better level of performance than rule- and statistics-based systems applied separately.

Human action recognition is used in areas such as surveillance, entertainment, and healthcare. This paper proposes a system to recognize both single and continuous human actions from monocular video sequences, based on 3D human modeling and cyclic hidden Markov models (CHMMs). First, for each frame in a monocular video sequence, the 3D coordinates of joints belonging to a human object, through actions of multiple cycles, are extracted using 3D human modeling techniques. The 3D coordinates are then converted into a set of geometrical relational features (GRFs) for dimensionality reduction and discrimination increase. For further dimensionality reduction, k-means clustering is applied to the GRFs to generate clustered feature vectors. These vectors are used to train CHMMs separately for different types of actions, based on the Baum-Welch re-estimation algorithm. For recognition of continuous actions that are concatenated from several distinct types of actions, a designed graphical model is used to systematically concatenate different separately trained CHMMs. The experimental results show the effective performance of our proposed system in both single and continuous action recognition problems.

In this paper, a novel vision-based nighttime vehicle detection approach is presented, combining state machines and downhill simplex optimization. In the proposed approach, vehicle detection is modeled as a sequential state transition problem; that is, vehicle arrival, moving, and departure at a chosen detection area. More specifically, the number of bright pixels and their differences, in a chosen area of interest, are calculated and fed into the proposed state machine to detect vehicles. After a vehicle is detected, the location of the headlights is determined using the downhill simplex method. In the proposed optimization process, various headlights were evaluated for possible headlight positions on the detected vehicles; allowing for an optimal headlight position to be located. Simulation results were provided to show the robustness of the proposed approach for nighttime vehicle and headlight detection.

Recent advances in positioning and photogrammetry technologies have made it possible to build a mobile mapping system (MMS) that can obtain 3D coordinates of geographic objects from stereo images recorded from a moving vehicle. In this paper, we present a design and detailed implementation of an MMS called a 4S-Van. Furthermore, we present four issues that have made major contributions to the performance of an MMS: 1) CCD camera calibration, 2) GPS signal condition, 3) integrating a GPS, inertial navigation system (INS), distance measurement indicator (DMI), and CCD cameras, and 4) the orientation angle of CCD cameras. In the experimental results, the performance of an MMS was analyzed for each component, giving an idea of how to effectively design and integrate each component in developing an MMS to get a maximal accuracy of 3D coordinates.

This letter presents a multi-modal approach to tracking geographic objects such as buildings and road signs in a video sequence recorded from a moving vehicle. In the proposed approach, photogrammetric techniques are successfully combined with conventional tracking methods. More specifically, photogrammetry combined with positioning technologies is used to obtain 3-D coordinates of chosen geographic objects, providing a search area for conventional feature trackers. In addition, we present an adaptive window decision scheme based on the distance between chosen objects and a moving vehicle. Experimental results are provided to show the robustness of the proposed approach.

In this paper, we present a probabilistic approach to determining whether extracted facial features from a video sequence are appropriate for creating a 3D face model. In our approach, the distance between two feature points selected from the MPEG-4 facial object is defined as a random variable for each node of a probability network. To avoid generating an unnatural or non-realistic 3D face model, automatically extracted 2D facial features from a video sequence are fed into the proposed probabilistic network before a corresponding 3D face model is built. Simulation results show that the proposed probabilistic network can be used as a quality control agent to verify the correctness of extracted facial features.

A novel idea is presented to locate intersections in a video sequence captured from a moving vehicle. More specifically, we propose a Bayesian network approach to combine evidence extracted from a video sequence and evidence from a database, maximizing evidence from various sensors in a systematic manner and locating intersections robustly. the causal relationship between events happening at intersections. Second, we construct a BN based on the analyzed causal relationship to successfully combine feature evidence extracted from a video sequence and evidence from a database to locate intersections robustly.