C # Research on touch screen calibration algorithm

 

The so-called touch screen calibration is actually the process of converting hardware coordinates to screen coordinates. Recently I tried to calibrate windows 7, but the general feeling is OK. However, there is a difference in windows 7 calibration, that is, the single-point touch screen and multi-point touch screen seem to need different calibration points, multi-point touch screens require 16-point calibration, and do not know whether the two methods adopt the same algorithm.

 

According to the needs of the product, we have recently studied various calibration algorithms for touch screens:

 

1. based on the linear proportional relationship between origins, the linear equation y = kx (k is the proportional coefficient) and k = screen horizontal (or vertical) Resolution/hardware horizontal resolution (or vertical) are established ); this method is simple, but its disadvantages are also obvious. This requires that the installation of hardware correspond to the screen one by one. In fact, there is no error in the installation process of hardware.

 

2. Through a general linear equation

 

A1 * x + B1 * y + C1 = Xd;

 

A2 * x + B2 * y + C2 = Yd;

 

X and y are the sample coordinates, and xd and yd are the corresponding screen coordinates.

 

To obtain relatively precise coefficients A1, B1, C1 ...... It needs to be converted into a quadratic equation, and then solved using the Gaussian elimination element or the gram diagonal line rule;

 

This algorithm seems simple, but it requires relatively complicated calculation to obtain a relatively accurate coefficient. However, this algorithm also has advantages and can use any number of calibration points, the number of calibration points must be determined based on the actual situation. Otherwise, too many calibration points will lead to redundancy. The calibration algorithm in Win CE seems to be using this method. The calibration result of this algorithm is good and suitable for touch screens.

 

3. rectangular transformation is actually using a digital image processing technology similar to that of the hardware. The hardware touch rectangle is converted into a screen-resolution rectangle through the affine transformation. This algorithm requires mathematical knowledge and is not described in detail.

 

This algorithm is suitable for touch screen and whiteboard calibration.

 

4. the method of calibration based on the Laplace interpolation algorithm is shown in the journal papers. It is found that the deviation of the calibration result is huge During the simulation test, I don't know if there is a bug in writing this algorithm on my side or it is not suitable for calibration itself. This algorithm is based on the least squares, and the least squares are mostly used for curve relations. I know more about it.

 

 

 

To sum up, except for the simplest one, I have used C # to implement other algorithms and use them on real Touch devices and electronic whiteboards. Based on the test results 2 and 3, the calibration results are good and have been applied to products. The above is my research. Please advise if you have not written the correct information, or if you have a better way to communicate with each other.

Author e @ Board