Research On The Method Of Close-range Large Field Central Projection Imaging Based On Rotating Camera

With the development of digital imaging, computer vision and digital image processing technology, digital photogrammetry has gotten rapid progress. The application of digital imaging technology has brought some convenience in digital photogrammetry, but it also brings the shortage that the FOV of digital camera is too smal, and it can guarantee the image overlap and make the width of the photo base as far as possible. which puts forward higher requirements when it is necessary to ensure the degree of overlap and to enlarge the baseline width in photogrammetry. To solve this contradiction, the method for increasing the viewing angle has been studied in aviation, aerospace photogrammetry, but there is few research and application in closerange photogrammetry. In order to obtain large format digital images in close-range photogrammetry, this paper which based on the rotating platform, studies the related theories and methods for acquiring wide-format images to satisfy the needs of photogrammetry. The main research contents and results are as follow:(1) Multi-camera stitching imaging, linear array scanning imaging and rotating multi-baseline imaging are the three methods to increase the viewing angle. By analyzing their principles and applications, the paper summarizes their limitations when applied to close-range photogrammetry and analyzes rotating cameras’ feasibility for close-range photogrammetry. Based on the collinear equation, the basic principles of rotating camera imaging are studied and influencing factors for imaging are analyzed.(2) In this paper, the image distortion caused by the camera rotating shooting is analyzed and the projection transformation model of rotating camera is established according to the PTZ initial attitude angle and the distortion of camera rotation is corrected. According to the principle, that the image points coordinate in the overlap area are same after projection transformation, the same name points of overlap area are obtained by SIFT matching to establish the relative error equation and to solve the exact value of the rotation angle. The accurate relative orientation and projection transformation of the image is realized to generate the virtual central projection large format image.(3) The image acquisition device is designed by camera, PTZ, controller and operation panel, and the real-time control and display of PTZ is realized. It is proved that the feasibility to obtain large format virtual image by rotary camera photogrammetry through two sets of experiments, and it’s verified that this method can achieve the accuracy requirements from the aspects of alignment errors and it can effectively increase the single lens digital camera image format.