Ipi mocap studio frames
exploit the iterative closest point (ICP) approach with a body model to track a human pose initialized by a hashing method. In model-based approaches, prior knowledge of a human body model is required, and the human pose is estimated by inverting the kinematics or solving optimization problems. Human pose estimation methods can be classified into two categories: model-based and learning-based approaches. Moreover, model-based approaches require model calibration before pose estimation. Other methods that do not use GPUs show low frame rates, and some cannot even run in real time. These methods shows remarkable performance, but it is difficult to operate the algorithms on low-cost systems, such as embedded boards or mobile platforms. Many human pose estimation methods use a GPU (graphic processing unit) to increase the frame rate and the performance. However, these sensors are limited to indoor use, and their low resolution and noisy depth information make it difficult to estimate human poses from depth images. The structured light sensors infer the depth values by projecting an infrared light pattern onto a scene and analyzing the distortion of the projected light pattern. Kinect for Xbox 360 and Xtion are RGB-D (red, green, blue and depth) sensors that obtain depth information by structured light technology. The release of low-cost depth sensors, such as Microsoft Kinect for Xbox 360 and ASUS Xtion, has provided many important benefits to these research areas. Human pose estimation and gesture recognition are attractive research topics in computer vision and robotics owing to their many applications, including human computer interaction, game control and surveillance. The experiment results show that our method performs fairly well and is applicable in real environments. We evaluated our method using a dataset that we generated. To prevent recognition of gestures when a person performs a gesture that is not registered, we derive the maximum allowable comparison errors by comparing each registered gesture with the other gestures. The gesture yielding the smallest comparison error is chosen as a recognized gesture. Gesture recognition is performed by comparing input keyframes with keyframes in registered gestures. To recognize gestures regardless of motion speed, the proposed method utilizes the keyframe extraction method. In the gesture recognition method, gestures are recognized from the pose information of a human body. The human pose estimation method is based on an SVM (support vector machine) and superpixels without prior knowledge of a human body model. The proposed methods are designed to be operated with only a CPU (central processing unit), so that the algorithm can be operated on a low-cost platform, such as an embedded board. In this paper, we present human pose estimation and gesture recognition algorithms that use only depth information.