Youngwoo Yoon

Blob extraction based character segmentation method for automatic license plate recognition system

A character segmentation algorithm for automatic license plate recognition is presented in this paper. Character regions are selected through binarization, connected component analysis, and character recognition. A blob analysis operation excludes noisy blobs, merges fragmented blobs, and splits clumped blobs. A character segmentation module achieved an accuracy rate of 97.2%. The recognition accuracy of the complete system with license plate localization was 90.9%. In depth analysis of failure cases is also provided for better understanding of the algorithm and a future development direction.

Real-Time Visual Target Tracking in RGB-D Data for Person-Following Robots

This paper describes a novel RGB-D-based visual target tracking method for person-following robots. We enhance a single-object tracker, which combines RGB and depth information, by exploiting two different types of distracters. First set of distracters includes objects existing near-by the target, and the other set is for objects looking similar to the target. The proposed algorithm reduces tracking drifts and wrong target re-identification by exploiting the distracters. Experiments on real-world video sequences demonstrating a person-following problem show a significant improvement over the method without tracking distracters and state-of-the-art RGB-based trackers. A mobile robot following a person is tested in real environment.

Blob detection and filtering for character segmentation of license plates

This paper presents a character segmentation method to address automatic number plate recognition problem. The method considered pixel intensity, character appearance, and arrangement of characters altogether to segment character regions. The method firstly discovers candidate blobs of characters by using connected component analysis and appearance-based character detection. A character recognizer is used for removing redundant and noisy blobs. Then, a trained classifier selects character blobs among the candidates by examining arrangement of the blobs. Experimental results show an achievement of 98.3% of segmentation rate, which prove the effectiveness of our method.

Depth assisted person following robots

This paper presents a person following robot equipped a RGB-D imaging sensor. We introduce three modules of a visual target tracking, target detection, and robot control. For the tracking module, distracters existing near-by the target are explicitly tracked to support target tracking. Preliminary tests showed that the robot robustly follows the target in uncontrolled environment with cluttered background and uneven illumination.

Number detection in natural image with boosting classifier

The number detection is useful in various applications such as license plate localization, detection of number button in elevator, and detection of exit number sign in public transport station. In this paper, we propose number detection methods in natural image using AdaBoost based on Modified Census Transform (MCT) features. It is a difficult task to detect numbers, characters, and specific symbols, because natural image includes many noises. Especially, illumination change is one of the most annoying sources of noise in the field of number detection based on image processing. Our number detection method uses many MCT features, which are robust to illumination change and AdaBoost for the feature selection to overcome this restriction. Experimental results show that the proposed method has a high detection rate in our license plate database which has been captured in the natural environment.

An experimental comparison of preprocessing methods for age classification

This paper investigates image transformations as preprocessing steps that can be applied toward a state-of-the-art age classification framework of manifold learning. We report on the experimental results of four different preprocessing methods in terms of classification accuracy using a large training and test face database. The use of histograms of oriented gradient (HOG) descriptors increases the classification accuracy from 46 to 75%. Robustness to the rotation and translation is also tested.

Person following with a RGB-D camera for mobile robots

This video demonstrates a person following robot that uses one RGB-D camera. A depth assisted color tracking method robustly tracks a fast moving target. The robot detects front obstacles from a depth image and avoids them. A person detector also works when the target is lost by occlusion and obstacle avoiding.

Classification of Dance Motions with Depth Cameras Using Subsequence Dynamic Time Warping

This paper proposes a method for classifying 3D dance motions especially selected from Korean POP (K-POP) dance performance, which is a key technique for the dance coaching contents and choreography retrieval system. Compared to actions addressed in daily life and existing games, K-POP dance motions are much more dynamic and vary substantially according to the performers. To cope with the variation of the amplitude of pose, we present a practical pose descriptor based on relative rotations between two body joints in the spherical coordinate system. As a method to measure similarity between two incomplete motion sequences, subsequence Dynamic Time Warping (DTW) algorithm is explored that supports partial matches. For the tests, 200 popular dance segments are gathered from 100 K-POP songs by utilizing the Kinect for Windows v2 sensor of Microsoft. The experimental results show that our representation and matching method can achieve an excellent performance in the classification of complex dance motions.

Choreography Retrieval from the Korean POP Dance Motion Capture Database with Low-Cost Depth Cameras

This paper proposes a method for searching a target choreography fraction from a motion capture database of the Korean POP (K-POP) dance. The proposed retrieval system allows users to create their own query sequences by performing dance with low-cost depth cameras. This intuitive search interface is essential for a retrieval of K-POP dance motions that have no official names for unit motions. As a method to describe and measure complex and dynamic dance poses, we utilize a relative angles between joints of interest. For speed up of matching motions, the two-phase approach is proposed which involves fast selection of candidates with key poses and precise comparison between motion segments by using Dynamic Time Warping method. The experimental results on a large database demonstrate that the performance of the system is a sufficiently practical level for real-world applications.

A perception framework for supporting robots to recognize human better in Human-Robot interaction

This paper describes the reason that perception technology in HRI has not given high performance enough to be used for commercial service robots. As a practical solution for better performance of perception technology, we propose a perception framework with a dedicated perception engine called a perception demon. The demon in the proposed framework constantly collects evidences and analyses them better by combining various types of individual perception components. The proposed framework enables robot makers to easily get more reliable information on humans without concerns about optimizing perception components to their robots.