Real-time depth-based hand detection and tracking.

Sung-Il Joo,Hyung-Il Choi,Sun-Hee Weon

doi:10.1155/2014/284827

Abstract

This paper illustrates the hand detection and tracking method that operates in real time on depth data. To detect a hand region, we propose the classifier that combines a boosting and a cascade structure. The classifier uses the features of depth-difference at the stage of detection as well as learning. The features of each candidate segment are to be computed by subtracting the averages of depth values of subblocks from the central depth value of the segment. The features are selectively employed according to their discriminating power when constructing the classifier. To predict a hand region in a successive frame, a seed point in the next frame is to be determined. Starting from the seed point, a region growing scheme is applied to obtain a hand region. To determine the central point of a hand, we propose the so-called Depth Adaptive Mean Shift algorithm. DAM-Shift is a variant of CAM-Shift (Bradski, 1998), where the size of the search disk varies according to the depth of a hand. We have evaluated the proposed hand detection and tracking algorithm by comparing it against the existing AdaBoost (Friedman et al., 2000) qualitatively and quantitatively. We have analyzed the tracking accuracy through performance tests in various situations.

Highlights

In the past decade, there have been intensive studies on the automatic analyses of human behaviors
Most of the red quadrangles are located within the hand region, which implies that the difference between the center and the average quadrangle region is larger than the threshold
In the case of number 10 shown in Figure 12, the red quadrangle is located outside a hand region, but it works for detection because of the small threshold value, −6231.1

Summary

Introduction

There have been intensive studies on the automatic analyses of human behaviors. The outliers are deleted by k-means clustering to detect the center point of the hand region, which becomes the base point for hand region detection and pose recognition This method, with a combined use of color and depth data, could improve precision in the detection process by removing outliers during filtering and applying a clustering technique. The Gaussian mixture model is used for calculating the probability distribution of the 3D x-coordinates and to detect the hand and the forearm regions This method detects a static pose, but it is limited when used for dynamic gesture recognition because the distribution model needs to be revised when the depth data change.

Hand Detection

Hand Tracking

Experimental Results

Conclusions