Browsing by Author "Kao, Chih-Chung"

Now showing 1 - 9 of 9

A New Approach to Video-Based Traffic Surveillance Using Fuzzy Hybrid Information Inference Mechanism
(2013-03-01)
Wu, Bing-Fei
;
Kao, Chih-Chung
;
Juang, Jhy-Hong
;
Huang, Yi-Shiun
;
電控工程研究所
This study proposes a new approach to video-based traffic surveillance using a fuzzy hybrid information inference mechanism (FHIIM). The three major contributions of the proposed approach are background updating, vehicle detection with block-based segmentation, and vehicle tracking with error compensation. During background updating, small-range updating is adopted to overcome environmental changes under congested conditions. During vehicle detection, the proposed approach detects the vehicle candidates from the foreground image, and it resolves problems such as headlight effects. The tracking technique is employed to track vehicles in consecutive frames. First, the method detects edge features in congested scenes. Next, FHIIM is employed to determine the tracked vehicles. Finally, a method that compensates for error cases under congested conditions is applied to refine the tracking qualities. In our experiments, we tested scenarios both inside and outside the tunnel with three lanes. The results showed that the proposed system exhibits good performance under congested conditions.
A Relative-Discriminative-Histogram-of-Oriented-Gradients-Based Particle Filter Approach to Vehicle Occlusion Handling and Tracking
(2014-08-01)
Wu, Bing-Fei
;
Kao, Chih-Chung
;
Jen, Cheng-Lung
;
Li, Yen-Feng
;
Chen, Ying-Han
This paper presents a relative discriminative histogram of oriented gradients (HOG) (RDHOG)-based particle filter (RDHOGPF) approach to traffic surveillance with occlusion handling. Based on the conventional HOG, an extension known as RDHOG is proposed, which enhances the descriptive ability of the central block and the surrounding blocks. RDHOGPF can be used to predict and update the positions of vehicles in continuous video sequences. RDHOG was integrated with the particle filter framework in order to improve the tracking robustness and accuracy. To resolve multiobject tracking problems, a partial occlusion handling approach is addressed, based on the reduction of the particle weights within the occluded region. Using the proposed procedure, the predicted trajectory is closer to that of the real rigid body. The proposed RDHOGPF can determine the target by using the feature descriptor correctly, and it overcomes the drift problem by updating in low-contrast and very bright situations. An empirical evaluation is performed inside a tunnel and on a real road. The test videos include low viewing angles in the tunnel, low-contrast and bright situations, and partial and full occlusions. The experimental results demonstrate that the detection ratio and precision of RDHOGPF both exceed 90%.
ACTIVE APPEARANCE MODEL ALGORITHM WITH K-NEAREST NEIGHBOR CLASSIFIER FOR FACE POSE ESTIMATION
(2014-06-01)
Wu, Bing-Fei
;
Kao, Chih-Chung
;
Jen, Cheng-Lung
;
Chiang, Chia-Rong
;
Lai, Po-Hung
In this paper, a face pose estimation (FPE) algorithm using active appearance model (AAM) with a k-nearest neighbor (kNN) classifier is presented. AAM is a model-based image representation method used to describe non-rigid visual objects, with both shape and texture variations, using a mean vector and linear combinations of a set of variation modes. Since AAM is a deformable model, it has several variations. Owing to the variations, the model is adjusted to the input test face image using iterative searching and fitting. The error, which measures the difference between the model and a test image, is minimized with the proposed searching algorithm. The face pose is then estimated using the distances between the landmark points in the AAM model with a INN classifier. Experimental results demonstrate that the proposed FPE algorithm can fit the face location with different face poses, with or without a hat, even wearing glasses, and identify the face pose accurately.
Adaptive feature selection for real-time face recognition in portable surveillance systems
(2007)
Kao, Wen-Chung
;
Shen, Chia-Ping
;
Kao, Chih-Chung
;
Hsu, Ming-Chai
;
Wu, Hung-Hsin
Real-time face recognition is a necessary feature in advanced portable surveillance systems. However, the high complexity of available algorithms to objects segmentation and recognition makes it impossible to include such a feature into a portable device. In this paper, we aim at designing a portable surveillance system which can take MPEG audio/video currently with recognizing human faces based on a digital camera platform. The proposed flow fully utilizes the available intermediate data passing from standard video compression flow in a digital camera. An efficient face recognition approach based on adaptive feature extraction and support vector machines (SVMs) are proposed to address the performance issues. The experimental result shows that the recognition speed running on a commercial digital camera platform can achieve 1.216 seconds/frame.
Embedded Weather Adaptive Lane and Vehicle Detection System
(2008)
Wu, Bing-Fei
;
Chen, Chao-Jung
;
Kao, Chih-Chung
;
Chang, Chia-Wei
;
Chiu, Shen-Ting
In this paper, the lane and vehicle detection with distance estimation algorithm is proposed by using a CCD camera mounted behind the windshield of our experimental car, TAIWAN iTS-1. The gray level and gradient value are applied for the lane marking detection. The lane marking information is utilized in vehicle detection algorithm. The front vehicles are recognized by comparing the gray level value and Sobel edge pixels. Then, the image coordinate model is utilized for the distance estimation with the located approaching vehicles. The verified effective distance is 50m in front of the host vehicle. This approach has been successfully implemented on the DSP system which is designed by us for the real-time detection. This system will be integrated into the lane departure warning and collision avoidance system with the visual and audio alerts to warn the careless drivers.
GPS TRACKING SYSTEM
(2010-12-23)
WU, BING-FEI
;
Chen, Ying-Han
;
Chen, Chao-Jung
;
Kao, Chih-Chung
;
Huang, Po-Chia
The present invention relates a GPS tracking system, comprising a tracking and positioning device, a server, a positioning display unit, and a database. Since the tracking and positioning device only sends an Internet protocol (IP) data and a verification data once when it is connected to the Internet, and sends a positioning data only when a position of the tracking and positioning device is to be shown on the positioning display unit, the bandwidth can be saved.
Method for image processing
(2010-06-24)
Wu, Bing-Fei
;
Chen, Chao-Jung
;
Kao, Chih-Chung
;
Chung, Meng-Liang
;
Chiu, Chung-Cheng
The invention relates to a method for image processing. First, establish the initial image background information. And retrieve the instant image information. Then calculate the initial image background information and color intensity information of the instant image. Furthermore, adjust the instant image information. Then calculate the moving-object information. Finally, track the moving-object information. It can improve the accuracy rate of detection without the influence of erected height.
Method for tracking and processing image
(2010-06-24)
Wu, Bing-Fei
;
Chen, Chao-Jung
;
Kao, Chih-Chung
;
Chung, Meng-Liang
;
Chiu, Chung-Cheng
The invention relates to a method for image processing, which can be used to calibrate the background quickly. When the external environment is changed due to the switch of light, the color of background is calibrated quickly, and the background can be updated together. The method not only is used to update the background, but also can be used to eliminate the convergence of background again.
The Vision-based Vehicle Detection and Incident Detection System in Hsueh-Shan Tunnel
(2008)
Wu, Bing-Fei
;
Kao, Chih-Chung
;
Liu, Chih-Chun
;
Fan, Chung-Jui
;
Chen, Chao-Jung
The automatic lane marking detection, vehicle detection and incident detection systems are proposed in this paper. The block-based background extraction that combines statistical algorithm and the moving block information is used to obtain the color background image more exactly. The lane detection algorithm is applied to obtain the lane information from the color background image without the limitation of the camera setting. The presented vehicle detection algorithm is a block-based approach which is widely used in highway systems, urban roadways or tunnels. Different applications adopt the different detection zones where the vehicles will be tracked and identified in. The presented incident detection algorithm focuses on the car stopped, vehicle lane changing, and congestion condition, and the experimental results in Hsueh-Shan tunnel are addressed, which also demonstrate the stability and the effectiveness of the proposed methods.