A simple pose estimation method based on co-planar feature points

In this paper, we propose cylindrical coordinate system which can analyze upper-body pose of the depth images correctly. This method extracts the part of human body from the depth images, and we configure the center of the part as origin of the cylindrical coordinate system. And we configure multi-layered cylinders based on the origin, then, we can analyze pose using the pattern which is crossed depth images namely cylinders and upper-body. Since the crossed point of the cylinders and upper-body is obtained as brightness values, it can extract to convert feature vector of the cylindrical coordinate system. The extracted feature vectors of the cylindrical coordinate system are presented to feature space of circular and are categorized pose patterns. The pose patterns are learned using average value of the feature vectors, and the pose patterns are categorized as pose to compare to pre-defined pose patterns using Euclidean distance. In this paper, we applied dynamic cylinder model to the region of the upper-body, so we can be classified as head, body and arms through simple computation, and to extract pose information is possible effectively. In this paper, the effect that can get through proposing pose estimation method is as following. At the first, pose estimation is possible by using only minimum feature points which apply cylinder model in region that connect human's torso, head and arms. The second is as following. When we obtain the feature points, because of applying cylinder model, we can extract user's feature points and angle of rotation easily. And in this paper, we don't consider the status of the user's body is titled using only the upper-body poses of the state rightly standing pose toward the front. Because it has disadvantage which cannot distinguish between changes according to the tile of the torso, but we can detect the vectors of the hands and arm using cylindrical coordinate system. Therefore, in the future, we will study to be able to recognize the pose in the tilted status.

Download Full-text

A Study of 3D Objects Pose Estimation for CAT System

Applied Mechanics and Materials ◽

10.4028/www.scientific.net/amm.36.485 ◽

2010 ◽

Vol 36 ◽

pp. 485-493

Author(s):

Abu Bakar Elmi ◽

Tetsuo Miyake ◽

Shinya Naito ◽

Takashi Imamura ◽

Zhong Zhang

Keyword(s):

Pose Estimation ◽

Estimation Method ◽

Feature Points ◽

Block Diagonalization ◽

Production Lines ◽

Method Performance ◽

3D Object ◽

3D Objects ◽

Model Based ◽

Object Feature

In production line, pose estimation of 3D object of products is needed beforehand. In order to perform shape measurement of the objects corresponding to speed of the mass production lines before the contact measurement is done, the information of object pose and matching is become required. In this paper, we conducted a study on the performance of model based and view based pose estimation method using image sequence of a rotating 3D object. In model based, we used object feature points from center of gravity and in view based method, the subspace calculation by block diagonalization of matrix represents a transformation an image to another image. We have confirmed the both method performance and it’s considered useful for pose estimation.

Download Full-text

3D Object’s Pose Estimation Based on Colored Markers Information

Advanced Materials Research ◽

10.4028/www.scientific.net/amr.846-847.1162 ◽

2013 ◽

Vol 846-847 ◽

pp. 1162-1165

Author(s):

Xiang Gao ◽

Chong Zhang ◽

Chun Gang Zhang ◽

Xi Juan Guo

Keyword(s):

Computer Vision ◽

Pose Estimation ◽

Color Space ◽

Estimation Method ◽

Regions Of Interest ◽

Experimental Results ◽

Feature Points ◽

Hsv Color Space ◽

Rigid Objects ◽

The Stability

Pose estimation of 3d object is a hot research in the field of computer vision. This paper presents a novel pose estimation method based on colored markers. To overcome the effect of the luminance and other colors, this method uses the HSV color space and isolates the colors operating only on chromaticity plane where value (V) has no actual effect for identifying the colored regions of interest. The template is then applied on the remaining colors in order to find the center of the region. The pixels which have the same color but are not in the marker area are excluded, since they are considered noisy. The template guarantees the stability and efficiency of the extraction of the feature points. Compared with the CDT algorithm, the proposed method can extract reliable center points, and has higher accuracy in pose estimation for planar rigid objects. At last, experimental results demonstrate the efficiency of the method.

Download Full-text

Real-time 6D Racket Pose Estimation and Classification for Table Tennis Robots

International Journal of Robotic Computing ◽

10.35708/rc1868-126249 ◽

2019 ◽

pp. 23-39

Author(s):

Yapeng Gao

Keyword(s):

Neural Network ◽

Real Time ◽

Pose Estimation ◽

Stereo Matching ◽

Estimation Method ◽

Average Error ◽

Feature Points ◽

Table Tennis ◽

Pose Detection ◽

Position And Orientation

For table tennis robots, it is a significant challenge to understand the opponent's movements and return the ball accordingly with high performance. One has to cope with various ball speeds and spins resulting from different stroke types. In this paper, we propose a real-time 6D racket pose detection method and classify racket movements into five stroke categories with a neural network. By using two monocular cameras, we can extract the racket's contours and choose some special points as feature points in image coordinates. With the 3D geometrical information of a racket, a wide baseline stereo matching method is proposed to find the corresponding feature points and compute the 3D position and orientation of the racket by triangulation and plane fitting. Then, a Kalman filter is adopted to track the racket pose, and a multilayer perceptron (MLP) neural network is used to classify the pose movements. We conduct two experiments to evaluate the accuracy of racket pose detection and classification, in which the average error in position and orientation is around 7.8 mm and 7.2 by comparing with the ground truth from a KUKA robot. The classification accuracy is 98%, the same as the human pose estimation method with Convolutional Pose Machines (CPMs).

Download Full-text

Modifications to Pose Estimation Algorithms that Allow for Near-Planar Feature Points

AIAA Guidance, Navigation and Control Conference and Exhibit ◽

10.2514/6.2008-7478 ◽

2008 ◽

Cited By ~ 2

Author(s):

Samuel Schweighart ◽

Piero Miotto ◽

Matthew Hale ◽

Leena Singh

Keyword(s):

Pose Estimation ◽

Feature Points ◽

Estimation Algorithms ◽

Planar Feature

Download Full-text

Optimal Pose Estimation Method for a Multi-Segment, Programmable Bevel-Tip Steerable Needle

2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) ◽

10.1109/iros45743.2020.9340847 ◽

2020 ◽

Author(s):

Alberto Favaro ◽

Riccardo Secoli ◽

Ferdinando Rodriguez y Baena ◽

Elena De Momi

Keyword(s):

Pose Estimation ◽

Estimation Method ◽

Steerable Needle

Download Full-text

A New Multi-Person Pose Estimation Method Using the Partitioned CenterPose Network

Applied Sciences ◽

10.3390/app11094241 ◽

2021 ◽

Vol 11 (9) ◽

pp. 4241

Author(s):

Jiahua Wu ◽

Hyo Jong Lee

Keyword(s):

Pose Estimation ◽

Human Body ◽

State Of The Art ◽

Estimation Method ◽

Bottom Up ◽

Center Point ◽

Novel Approach ◽

Body Joints

In bottom-up multi-person pose estimation, grouping joint candidates into the appropriately structured corresponding instance of a person is challenging. In this paper, a new bottom-up method, the Partitioned CenterPose (PCP) Network, is proposed to better cluster the detected joints. To achieve this goal, we propose a novel approach called Partition Pose Representation (PPR) which integrates the instance of a person and its body joints based on joint offset. PPR leverages information about the center of the human body and the offsets between that center point and the positions of the body’s joints to encode human poses accurately. To enhance the relationships between body joints, we divide the human body into five parts, and then, we generate a sub-PPR for each part. Based on this PPR, the PCP Network can detect people and their body joints simultaneously, then group all body joints according to joint offset. Moreover, an improved l1 loss is designed to more accurately measure joint offset. Using the COCO keypoints and CrowdPose datasets for testing, it was found that the performance of the proposed method is on par with that of existing state-of-the-art bottom-up methods in terms of accuracy and speed.

Download Full-text