US Patent No. 10,679,043

3-D PATH DETECTION SYSTEM FOR HAND GESTURE RECOGNITION APPLICATION


Patent No. 10,679,043
Issue Date June 09, 2020
Title 3-d Path Detection System For Hand Gesture Recognition Application
Inventorship Tzyy-Sheng Horng, Kaohsiung (TW)
Chia-Hung Yeh, Kaohsiung (TW)
Fu-Kang Wang, Kaohsiung (TW)
Mu-Cyun Tang, Kaohsiung (TW)
Chien-Lun Chen, Kaohsiung (TW)
Min-Hui Lin, Kaohsiung (TW)
Assignee NATIONAL SUN YAT-SEN UNIVERSITY, Kaohsiung (TW)

Claim of US Patent No. 10,679,043

1. A 3-D path detection system comprising:an image capture device configured to produce a dynamic image, wherein the dynamic image involves a moving object image, and the image capture device is configured to calculate an x-direction (transverse horizontal) pixel-value displacement and a y-direction (transverse vertical) pixel-value displacement of the moving object image;
a radar device configured to transmit an input wireless signal to a moving object, receive a reflection signal from the moving object and obtain a z-direction (longitudinal) displacement of the moving object according to a Doppler shift in the reflection signal; and
a computing module coupled to the image capture device and the radar device, wherein the computing module is configured to perform a calibration procedure to convert the x- and y-direction pixel-value displacements into a x-direction displacement and a y-direction displacement by relating to the z-direction displacement, and the computing module is configured to construct a 3-D path of the moving object by using the x- and z-direction displacements of the moving object;
wherein the calibration procedure is: the moving object moves along a straight path in the xz plane with a first angle relative to x-axis and then moves along a straight path in the yz plane with a second angle relative to y-axis, and the first and second angles are the predetermined angles, wherein according to the x- and y-direction pixel-value displacements of the moving object image and the z-direction displacement of the moving object, the computing module is configured to calculate a first conversion coefficient from the x-direction pixel-value displacement of the moving object image to an x-direction displacement of the moving object and calculate a second conversion coefficient from the y-direction pixel-value displacement of the moving object image to a y-direction displacement of the moving object, and the calculation formulas of the first and second conversion coefficients are represented as follows:

wherein mx is the first conversion coefficient, ?z is the z-direction displacement of the moving object, ?Px is the x-direction pixel-value displacement of the moving object image, ?x is the first angle, ?x is the x-direction displacement of the moving object, my is the second conversion coefficient, ?Py is the y-direction pixel-value displacement of the moving object image, ?y is the second angle, and ?y is the y-direction displacement of the moving object.