• Journal of Korea Multimedia Society
• /
• 제19권5호
• /
• pp.865-873
• /
• 2016

In recent years, Tracking-Learning-Detection(TLD) system has been widely used as a detection and tracking algorithm for vision sensors. While conventional algorithms are vulnerable to occlusion, and changes in illumination and appearances, TLD system is capable of robust tracking by conducting tracking, detection, and learning in real time. However, the detection and tracking algorithms of TLD system utilize rotation-variant features, and the margin of tracking error becomes greater when an object makes a full out-of-plane rotation. Thus, we propose a rotation-invariant TLD system(RI-TLD). we propose a simplified average orientation histogram and rotation matrix for a rotation inference algorithm. Experimental results with various tracking tests demonstrate the robustness and efficiency of the proposed system.

• Journal of the Korea Society of Computer and Information
• /
• 제26권11호
• /
• pp.85-92
• /
• 2021

In this paper, a method for effectively detecting rotated face and rotation angle regardless of the rotation angle is proposed. Rotated face detection is a challenging task, due to the large variation in facial appearance. In the proposed polar coordinate transformation, the spatial information of the facial components is maintained regardless of the rotation angle, so there is no variation in facial appearance due to rotation. Accordingly, features such as HOG, which are used for frontal face detection without rotation but have rotation-sensitive characteristics, can be effectively used in detecting rotated face. Only the training data in the frontal face is needed. The HOG feature obtained from the polar coordinate transformed images is learned using SVM and rotated faces are detected. Experiments on 3600 rotated face images show a rotation angle detection rate of 97.94%. Furthermore, the positions and rotation angles of the rotated faces are accurately detected from images with a background including multiple rotated faces.

• IEIE Transactions on Smart Processing and Computing
• /
• 제1권2호
• /
• pp.73-77
• /
• 2012

This paper reports a combined depth- and model-based face detection and tracking approach. The proposed algorithm consists of four functional modules; i) color-based candidate region extraction, ii) generation of the depth histogram for handling occlusion, iii) rotation-invariant face region detection using ellipse fitting, and iv) face tracking based on motion prediction. This technique solved the occlusion problem under complicated environment by detecting the face candidate region based on the depth-based histogram and skin colors. The angle of rotation was estimated by the ellipse fitting method in the detected candidate regions. The face region was finally determined by inversely rotating the candidate regions by the estimated angle using Haar-like features that were robustly trained robustly by the frontal face.

• Journal of the Korean Society for Nondestructive Testing
• /
• 제34권6호
• /
• pp.447-456
• /
• 2014

In this paper, a new damage-detection method based on structural vibration is proposed. The essence of the proposed method is the detection of abrupt changes in rotation. Damage-induced rotation (DIR), which is determined from the modal flexibility of the structure, initially occurs only at a specific damaged location. Therefore, damage can be localized by evaluating abrupt changes in rotation. We conducted numerical simulations of two damage scenarios using a 10-story cantilever-type building model. Measurement noise was also considered in the simulation. We compared the sensitivity of the proposed method to localize damage to that of two conventional modal-flexibility-based damage-detection methods, i.e., uniform load surface (ULS) and ULS curvature. The proposed method was able to localize damage in both damage scenarios for cantilever structures, but the conventional methods could not.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• 제39권5호
• /
• pp.535-543
• /
• 2002

In this paper, we propose a new gaze detection method using 2-D facial images captured by a camera on top of the monitor. We consider only the facial rotation and translation and not the eyes' movements. The proposed method computes the gaze point caused by the facial rotation and the amount of the facial translation respectively, and by combining these two the final gaze point on a monitor screen can be obtained. We detected the gaze point caused by the facial rotation by using a neural network(a multi-layered perceptron) whose inputs are the 2-D geometric changes of the facial features' points and estimated the amount of the facial translation by image processing algorithms in real time. Experimental results show that the gaze point detection accuracy between the computed positions and the real ones is about 2.11 inches in RMS error when the distance between the user and a 19-inch monitor is about 50~70cm. The processing time is about 0.7 second with a Pentium PC(233MHz) and 320${\times}$240 pixel-size images.

• Proceedings of the IEEK Conference
• /
• 대한전자공학회 2007년도 하계종합학술대회 논문집
• /
• pp.339-340
• /
• 2007

Face detection plays an important role in HCI and face recognition. In this paper, we propose a rotation-invariant real-time face detection algorithm for color images in complex background. It consists of four processing step: (1) motion detection, (2) skin color region filler, (3) Eyemap detector for rotated face, and (4) Adaboost face classifier. This system has been tested in in-door environments, such as office and achieves over 95% detection rate.

• The Journal of Korea Robotics Society
• /
• 제14권1호
• /
• pp.14-21
• /
• 2019

In the ground environment, mobile robot research uses sensors such as GPS and optical cameras to localize surrounding landmarks and to estimate the position of the robot. However, an underwater environment restricts the use of sensors such as optical cameras and GPS. Also, unlike the ground environment, it is difficult to make a continuous observation of landmarks for location estimation. So, in underwater research, artificial markers are installed to generate a strong and lasting landmark. When artificial markers are acquired with an underwater sonar sensor, different types of noise are caused in the underwater sonar image. This noise is one of the factors that reduces object detection performance. This paper aims to improve object detection performance through distortion and rotation augmentation of training data. Object detection is detected using a Faster R-CNN.

• Journal of Korea Multimedia Society
• /
• 제17권9호
• /
• pp.1070-1075
• /
• 2014

Due to a number of illegal copied videos, many video content markets have been threatened. Since this copied videos have intercepted the profits of the content holders, content developers lose the will to generate new contents. Therefore, video copy detection approaches have been developed to protect the copyrights of video contents. However, many illegal uploader who generate copied videos used video transformations to avoid video copy detection systems. Among of the video transformations, rotation and flipping did not distorted the quality of video contents. Thus, these two video transformations were adopt to generate copied video. In order to detect rotated or flipping copy videos, rotation and flipping robust region binary pattern (RFR) recently was proposed. But, this RFR has a weakness according to rotated angles. Therefore, in order to overcome this problem, multi region binary patterns are proposed in this paper. The proposed method has the similar performance with the original RFR. But, it showed much higher efficiency for memory spaces.

• Journal of the Korean Institute of Intelligent Systems
• /
• 제17권1호
• /
• pp.58-65
• /
• 2007

The method of 3D sound realization using 2 speakers provides two advantages: cheap and easy to build. In the case, crosstalk between 2 speakers has to be eliminated. To calculate and remove the effect of the crosstalk it is essential to find a rotation angle of human head correctly. In the paper, we suggest an algorithm to find the head angle of 2 channel system. We first detect a face area of the given image using Haar-like feature. After that, the eve detection using pre-processor and morphology method. Finally, we calculate the face rotation angle with the face andi the eye location. As a result of the experiment on various face images, the proposed method improves the efficiency much better than the conventional methods.

• International Journal of Internet, Broadcasting and Communication
• /
• 제9권1호
• /
• pp.64-70
• /
• 2017

In the commercial building such as huge enterprise building, more accurate operation of the center-controlled roller blind. We design, in this work, the target disc that its shape is nonlinearly changing and the sensor coils that are differentially arranged. The performance shows less than 1% accuracy when it is implemented in the roller blind.