Research On Geometric And Radiation Consistency Processing Method Of DMZ? Camera Image

With the rapid development of microelectronic technology and photogrammetry,the way to obtain aerial images has turned to digital aerial camera.Limited by the development technology of CCD,the current imaging system of large format aerial camera is usually composed of several medium-sized CCD,which makes the resolution of the obtained images and the coverage of the ground improved at the same time,and can realize the observation and positioning of the ground more efficiently.The data preprocessing of large-format aerial survey cameras has also become a research hotspot,and the research on the generation technology of central projection or equivalent central projection images has received the most general attention.This paper mainly focuses on the research on the image geometric radiation consistency processing method of DMZ?,a single-lens multi-area CCD aerial survey camera independently developed by our country.According to the imaging characteristics of the DMZ? camera,a panchromatic image geometry of the DMZ? camera is proposed.The consistency processing algorithm and radiation consistency processing algorithm have been tested to verify the effectiveness and applicability of the method.The main content and conclusions of this article include:1.This paper briefly introduces the technical system and imaging principle of typical digital aerial camera at home and abroad,and analyzes the innovation,radiation characteristics and geometric imaging principle of DMZ? camera,the data source of this paper.This paper describes the research status of image geometric consistency processing technology and radiation consistency processing technology at home and abroad,and introduces some basic theories used in this paper,including coordinate system commonly used in photogrammetry,matching algorithm,transformation model,resampling and overlapping region fusion algorithm involved in image geometric consistency processing.2.Aiming at the DMZ? aerial imagery,using the strategy of “correcting before stitching”,an equivalent large-format aerial image geometric consistency processing method is proposed.Firstly,12 sub images are processed by radiometric correction and geometric correction to get non distortion sub images,and then feature points are extracted between adjacent sub images according to the image overlapping area.The mismatched points are eliminated by using the algorithm based on image point coordinate offset and random sampling consistency algorithm,and then the high-precision homonymous points are obtained by using the least square image matching,Finally,according to the same poi nts,the mosaic parameters are calculated by using the translation transformation model,and the sub images are stitched to obtain the equivalent large-scale aerial image.3.Aiming at the situation that there are radiation differences at the edges of the subimage positions in the generated equivalent aerial images,a method for processing panchromatic image radiation consistency based on low-resolution multi-spectral images is proposed.After getting the equivalent large format image,the low-resolution multi-spectral image is registered with the panchromatic image to obtain a multi-spectral image of the same size.The multi-spectral image itself has good radiation consistency,and the radiation consistency processing is performed on the panchromatic image in the same measurement area.The experimental results prove that the algorithm can effectively eliminate the radiation difference of the image,and the image quality is improved.