Unmanned aerial vehicle scene stereo matching data generation method

The invention relates to an unmanned aerial vehicle scene stereo matching data generation method, and belongs to the technical field of unmanned aerial vehicle image stereo matching. According to the invention, combined modeling is carried out by using the image acquired by the unmanned aerial vehicle and the laser point cloud data, so that a three-dimensional model with fine texture, high authenticity and relatively high precision is reconstructed as a simulation scene of low-altitude aerial photography of the unmanned aerial vehicle; the method comprises the following steps: acquiring left and right view images of a stereo image pair by using a simulated binocular camera acquisition mode, and then acquiring the depth of each pixel point by using a model calculation method so as to obtain a corresponding parallax error and provide a dense parallax error true value for an actually measured image; and an image obtained by the unmanned aerial vehicle is used as an actual measurement image, and actual measuremen