| 摘要 |
A camera calibration method for image measurement is provided to carry out exact camera calibration even under unequal lighting conditions in an industrial site by automatically obtaining the optimum camera calibration parameters using inputted 4-point information, even though a lighting noise occurs at a pattern image. If a user inputs a calibration point pattern image, a camera calibration system executes preprocessing through binarization and labeling. The user selects 4 calibration points, and inputs the number of horizontal calibration points and the number of vertical calibration points, which are located between the 4 calibration points. The system executes planar perspective transformation with image coordinates and world coordinates for the 4 points and obtains a planar perspective transformation matrix "H". The system multiplies the coordinate value of a calibration point on the world coordinates by the obtained matrix, obtains the matrix of the image coordinates, and stores coordinates which are smaller than the first threshold. Using the stored world and image coordinates as the input values of an algorithm to calculate camera calibration, the system executes tsai calibration. Then the system obtains image coordinates through an operation for the world coordinates and internal/external parameters, obtains an inverse projection value using the image coordinates and camera parameters, calculates a calibration error, and checks whether the calculated error value is smaller than the second threshold.
|
