Distinctive Image Features from Scale-Invariant Keypoints
International Journal of Computer Vision
Efficient Maximally Stable Extremal Region (MSER) Tracking
CVPR '06 Proceedings of the 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Volume 1
Supervised particle filter for tracking 2D human pose in monocular video
WACV '11 Proceedings of the 2011 IEEE Workshop on Applications of Computer Vision (WACV)
Color-based extensions to MSERs
ISVC'11 Proceedings of the 7th international conference on Advances in visual computing - Volume Part II
SURF: speeded up robust features
ECCV'06 Proceedings of the 9th European conference on Computer Vision - Volume Part I
| Hi-index | 0.00 |
Human Skeletal Tracking has a variety of applications in many vision and graphics tasks like Gesture Recognition and 3D Motion Reconstruction. In this paper, we present a novel approach to human body skeletal tracking from monolocular Human motion video sequences. We use the colour MSER region detectors to identify and track the location of hands, head and feet to be used for skeletal tracking. Based on the position of detected end effectors we fit a skeleton for pose recovery. The proposed method does not involve any models or learning as is based on inter frame feature correspondence. The framework for pose recovery has been tested on human action videos taken from the Microsoft Kinect and has been found to yield good results. We also perform a quantitative evaluation of the feature detectors in terms of the number of useful features detected as well as the running time complexity which has shown Colour MSER features to perform well compared to other existing feature detectors.