Browsing by Author "Lee, Hsi-Jian"

Now showing 1 - 7 of 7

A Spatial-Extended Background Model for Moving Blobs Extraction in Indoor Environments
(2009-11-01)
Zhao, San-Lung
;
Lee, Hsi-Jian
;
資訊工程學系
;
Department of Computer Science
This paper presents a system for extracting regions of moving objects from an image sequence. To segment the foreground regions of ego-motion objects, we create a background model and update it using recent background variations. Since background images are usually changed in blobs, spatial relations are used to represent background appearances, which may be affected drastically by illumination changes and background object motion. To model the spatial relations, the joint colors of each pixel-pair are modeled as a mixture of Gaussian (MoG) distributions. Since modeling the colors of all pixel-pairs is expensive, the colors of pixel-pairs in a short distance are modeled. The pixel-pairs with higher mutual information are selected to represent the spatial relations in the background model. Experimental results show that the proposed method can efficiently detect the moving object regions when the background scene changes or the object moves around a region. By comparing with Gaussian background model and the MoG-based model, the proposed method can extract object regions more completely.
APPLICATION OF SPATIOTEMPORAL SEARCH-SPACE REDUCTION IN LICENSE PLATE DETECTION
(2008)
Wang, Shen-Zheng
;
Lee, Hsi-Jian
;
資訊工程學系
;
Department of Computer Science
In surveillance applications, search space reduction (SSR) is an essential element to efficient algorithms. In this study, spatial and temporal SSRs are integrated for license plate detection in video sequences; the plates could be extracted robustly and extremely fast. Our method started from spatial SSR. by a bi-level one-pass plate extraction (BOPE) algorithm developed to extract plates accurately even in complicated situations. Moreover, we proposed to exclude repeated patterns with the similar appearances in the same location of consecutive frames, which usually include stopped vehicles or regular backgrounds. For efficiency, repeated patterns were detected only on the results of BOPE, named spatiotemporal SSR, based on a block-based mechanism by estimating the tangent distance, which is invariant to the variations in positions, sizes, rotations, or brightness. To reduce the computational load, the repeated patterns and their measured invariant features could be retained for next estimation. In our experiments, the search space could be reduced up to 87.9% by the spatiotemporal SSR; all plate numbers with the heights ranging from 24 to 40 are extracted correctly. The high performance, average 76 frames per second on a 3G Hz PC, suggests that the real-time goal for surveillance applications could be accomplished.
Cascade plate recognition system
(2007-12-20)
Wang, Shen-Zheng
;
Lee, Hsi-Jian
The present invention discloses a cascade plate recognition system, wherein a compact plate region detection module receives an image, detects the plate area candidates of the image, picks out the sequences having continuous identical pixels from the plate area candidates, performs smearing and filtering on the sequences, extracts connected components from the plate area candidates to obtain compact plate regions, verifies the compact plate regions and outputs the verified compact plate regions; a plate character segmentation and recognition module receives the verified compact plate regions, extracts independent-character regions from each verified compact plate regions, verifies the independent-character regions and recognizes the verified independent-character regions to obtain the information of the plate characters. Thereby, the present invention has the advantages of high accuracy, learning capability and fast realtime operation.
Human silhouette extraction based on HMM
(2006)
Zhao, San-Lung
;
Lee, Hsi-Jian
;
資訊工程學系
;
Department of Computer Science
This paper presents a system that can extract regions of a person from an image sequence. The system first detects foreground regions based on a background model. After foreground regions are extracted a human model is used to identify human regions. In the human model, we adopt a Hidden Markov Model (HMM to model human silhouettes. To separate human and non-human regions, we represent non-human silhouettes by another HMM. The human silhouette model can extract incomplete human regions even when parts of the person are covered by background objects. Experimental results show the system proposed is very effective.
LICENSE PLATE RECOGNITION SYSTEM USING SPATIAL-TEMPORAL SEARCH-SPACE REDUCTION AND METHOD THEREOF
(2009-08-20)
Wang, Shen-Zheng
;
Lee, Hsi-Jian
The present invention discloses a license plate recognition system by utilizing spatial-temporal search-space reduction and method thereof. The license plate recognition system can selects a pixel with character edge feature from the inputted image and a plurality of candidate regions from the image; and detect if the candidate region of the inputted image has been appeared in a preceding image. Then the recognition system in the present invention separates the candidate region into a plurality of blocks, calculates the repeated block based on a current image and a preceding image, and determines whether the candidate region is repeated according to the repeated block. Thus, by discarding the repeated regions, the license plate recognition system in the present invention can avoid redundant calculation and improve the efficiency and performance for recognition of real plate character in the image.
Particle Filter-based Multi-part Human Tracking with Failure Adjustment in Video Sequences
(2010-11-01)
Zhao, San-Lung
;
Lee, Hsi-Jian
;
資訊工程學系
;
Department of Computer Science
The study presents a human tracking system in video sequences. To track the target, we. first detect humans in a video according to a Gaussian background model. We then track the humans by using color histogram as the features and using particle filters as the tracking kernel. Since a human is not a rigid object, his appearance may be greatly affected by his motion. In our applications, human bodies imaged are generally large. We decompose each human body into three parts: head, torso, and hip-leg, represent them by three shrunk rectangles, and track them by particle filters. In this way we can reduce possible tracking failures by checking the interrelationship among these three parts. We use support vector machines (SVM) to detect tracking failures and abnormal body parts since the abnormal situations are very diversified and cannot be easily encoded in rules. If a single part is abnormal, its position can be adjusted from the other parts and tracked using the system dynamic model. If two or three parts are abnormal, we re-initialize the tracking process of the three parts around their predicted positions. By testing on 22 video clips from six scenes, the experimental results showed that our three-part tracking system with failure detection and correction can track correctly about 95% persons until the 105th frame. With respect to the body parts, our system has about 95%, 83%, and 91% tracking rates for head, torso, and hip-leg respectively until the 105th frame. The tracking rate of a human increase 20% comparing with that of the whole-body tracker. These rates show the effectiveness of the proposed system.
Particle Filter-based Multi-part Human Tracking with Failure Adjustment in Video Sequences
(2010-11-01)
Zhao, San-Lung
;
Lee, Hsi-Jian
;
資訊工程學系
;
Department of Computer Science
The study presents a human tracking system in video sequences. To track the target, we. first detect humans in a video according to a Gaussian background model. We then track the humans by using color histogram as the features and using particle filters as the tracking kernel. Since a human is not a rigid object, his appearance may be greatly affected by his motion. In our applications, human bodies imaged are generally large. We decompose each human body into three parts: head, torso, and hip-leg, represent them by three shrunk rectangles, and track them by particle filters. In this way we can reduce possible tracking failures by checking the interrelationship among these three parts. We use support vector machines (SVM) to detect tracking failures and abnormal body parts since the abnormal situations are very diversified and cannot be easily encoded in rules. If a single part is abnormal, its position can be adjusted from the other parts and tracked using the system dynamic model. If two or three parts are abnormal, we re-initialize the tracking process of the three parts around their predicted positions. By testing on 22 video clips from six scenes, the experimental results showed that our three-part tracking system with failure detection and correction can track correctly about 95% persons until the 105th frame. With respect to the body parts, our system has about 95%, 83%, and 91% tracking rates for head, torso, and hip-leg respectively until the 105th frame. The tracking rate of a human increase 20% comparing with that of the whole-body tracker. These rates show the effectiveness of the proposed system.