Calibration method for virtual touch control

Abstract

This invention provides a virtual touch calibration method, comprising: an image capturing unit capturing a target area to generate an original image; performing a perspective transformation on the original image using a direct linear transformation algorithm to generate a perspective image; performing lens distortion calibration on the perspective image using extended precision floating-point numbers to generate a calibration image; performing depth parallax calibration on the calibration image using triangulation formulas to generate a depth image; calibrating the depth image using translation calibration to generate a translation image; the host computer calibrating the translation image using a nonlinear calibration method to generate a first image; compensating the first image using progressive compensation to generate a second image; filtering the second image using a jump-point filtering method to generate a filtered image; and performing edge enhancement on the filtered image using smoothing processing to generate a third image.

Claims

1. A virtual touch calibration method applied to a target area of ​​an object's occluded portion, the method comprising: an image capturing unit capturing the target area and generating an original image, and transmitting the original image to a host computer; the host computer performing a perspective transformation on the original image using a direct linear transformation algorithm and generating a perspective image; the host computer performing lens distortion calibration on the perspective image using an extended precision floating-point number and generating a calibration image; the host computer performing depth parallax calibration on the calibration image using a triangulation formula and generating a depth image; the host computer calibrating the depth image using a translation calibration and generating a translation image; the host computer calibrating the translation image using a nonlinear calibration method and generating a first image; the host computer compensating the first image using a progressive compensation and generating a second image; the host computer filtering the second image using a jump-point filtering method and generating a filtered image; the host computer performing edge enhancement on the filtered image using a smoothing process and generating a third image; and the host computer using the third image as an input command and executing a touch instruction.

2. The virtual touch calibration method as described in claim 1, wherein in the steps of capturing the target area and generating an original image in an image capturing unit and transmitting the original image to a host, the original image includes a plurality of touch positions.

3. The virtual touch calibration method as described in claim 1, wherein in the step of the host performing a perspective transformation on the original image using a direct linear transformation algorithm to generate a perspective image, the perspective transformation is calculated using singular value decomposition to obtain a transformation matrix.

4. The virtual touch calibration method as described in claim 1, wherein in the step of the host performing a lens distortion calibration on the perspective image using an extended precision floating-point number and generating a calibration image, the extended precision floating-point number corresponds to 128 bits.

5. The virtual touch calibration method as described in claim 1, wherein in the step of the host performing a depth parallax calibration on the calibration image using a triangulation formula and generating a depth image, the triangulation formula includes a baseline distance and a focal length, the baseline distance being 50 mm and the focal length being 191.575 pixels.

6. The virtual touch calibration method as described in claim 1, wherein in the step of the host calibrating the depth image using a translation calibration and generating a translation image, the translation calibration is performed using a piecewise linear interpolation method.

7. The virtual touch calibration method as described in claim 1, wherein in the step of the host calibrating the translation image using a nonlinear calibration method and generating a first image, the nonlinear calibration method is compensated using a quadratic function model.

8. The virtual touch calibration method as described in claim 1, wherein in the step of the host compensating the first image using a progressive compensation to generate a second image, the progressive compensation is performed using a compensation value that increases with the distance from the center of the first image.

9. The virtual touch calibration method as described in claim 1, wherein in the step of the host filtering the second image using a jump point filtering method and generating a filtered image, the jump point filtering method has a movement distance threshold and a movement speed threshold, and the second image is filtered when a movement distance of the second image is greater than or equal to the movement distance threshold and when a movement speed of the second image is greater than or equal to the movement speed threshold.

Images