FIELD: computing; counting.
SUBSTANCE: invention relates to the field of calibration of video cameras, working in the vision system. Method for calibrating video sensors of a multispectral vision system includes shooting a test object from different angles, frame conversion into digital form, finding the special points on the images taken and estimating their coordinates with sub-pixel accuracy, estimating the matrixes of internal camera parameters, estimating the vectors of camera lens distortion coefficients, estimating external parameter matrices and carry out the simultaneous calibration of cameras in the visible and infrared ranges with intersecting fields of view, a test object is used in the form of a heated heat-conducting calibration template with dark n-sided polygons, and after receiving half-tone images from multi-spectral cameras at different positions of the test object, the images from infrared cameras are inverted.
EFFECT: technical result is to obtain a high-contrast image of a test pattern observed by cameras of the visible and infrared range to calibrate video sensors of a multispectral vision system.
1 cl, 1 dwg
