The Study On Kinect Based Finger Joints Tracking Techniques

In recent years,with the hot development of Virtual Reality,many popular new technologies were derived,in which hand and finger tracking technology is a hot research topic in VR field in the last decade.The finger tracking has wide possibility for applications,such as in the field of human-computer interaction(HCI),computer animation,and video games,and people's interest in this topic is still growing.For the 3D HCI,real-time,stability and robustness are the main requirements of 3D finger joints tracking,only if these requirements are met,could people control virtual object just like touching object in reality.So,for good user experience,finger tracking algorithm need to be further improved.In this paper,we use Microsoft developed Kinect v2 somatosensory camera as depth image capture device,and proposed a depth based real-time robust and stable finger tracking method.Firstly,we use human skeleton recognition function in Kinect for Windows SDK to get the location of palm,and divide a hand region with the palm center.Then the region is denoised with the improved median filter,and from here,hand contour is extracted.The contour is used to calculate the convex hull of the hand contour by the modified Graham scanning method,and we can find finger tips from the convex hull.Then we can get other finger joints according to the morphological characteristics of the finger contour.At last,the movement of the finger joints is tracked according to the customized finger IK constraint model proposed in this paper.An optimized median filter algorithm and optimized Graham convex hull algorithm are used to process the depth information of the hand region more quickly.By using our method,the three nodes of each finger can be tracked in real time in the demand of natural HCI.Compared with the existing methods,our method has many advantages,and great application prospects in 3D HCI field.