Research On Constrained Camera Iterative Calibration And Visible-band Based Bathymetry Methods

The ocean not only contains a lot of rich resources but is also a channel connecting the major sectors of the world.It has a profound impact on the policies,economy,culture,and national security of all countries,and has a very important strategic position.With the progress of society and the development of science and technology,the role of the oceans is becoming greater and greater.my country is a large marine country with a coordinated land and sea,and there are many lakes in our country.The use and control of marine resources,the development of marine economic construction,and the development of lake resources are inseparable from water depth and bottom topography information.The traditional acquisition of water depth and bottom terrain information relies on ships and manpower to conduct field surveys,which is time-consuming and requires sufficient manpower.The increasing maturity of remote sensing technology makes it possible to obtain water depth and topographic information without field surveys.Remote sensing technology breaks through the limitations of time and space,can carry out high-efficiency and large-scale detection,and its spatial resolution is high.It is a very desirable and efficient technology to use remote sensing inversion to obtain water depth and bottom topography information of specific waters.means.Therefore,it is necessary to research the methods of obtaining water depth information using remote sensing images.The current two-medium stereo photogrammetry refraction correction method obtains the water depth information through the original three-dimensional image and obtains the real water depth information by refraction correction on the intersection point of the extension line of the incident light with the same name.The accuracy of the water depth information will be affected by the intersection point(the position relationship is uncertain).To avoid the influence of the accuracy of the theoretical intersection point on the accuracy of underwater objects,this paper introduces the depth refraction method based on projected images.This method is based on the principle of refraction correction in two-medium stereo photogrammetry,and the refracted rays are directly intersected to obtain water depth information.The inversion model commonly used in the multi-spectral remote sensing method of bathymetry requires sufficient measured water depth information for modeling,and then the water depth is inverted based on the trained inversion model.Last several years,with the rapid development of UAV low-altitude remote sensing technology,aerial photography becomes an important way to obtain aerial images for photogrammetry and remote sensing applications.However,in general the cameras used by UAV are not professional cameras for surveying and mapping purposes.Compared to the professional cameras,the non-metric cameras usually have large lens distortions,which will have severe impacts on the obtained spatial information accuracy.Therefore,it is necessary and critical to accurately calibrate non-metric cameras.After compensating the lens distortions,high-precision water depth information can then be obtained afterwards.The traditional non-metric camera calibration methods depend on the highprecision indoor or outdoor control fields,and have some restrict requirements for the calibration environments,this paper proposes a new camera calibration method that completely abandons any control points,only uses some well available features such as lines,planes or angles as constraints.In the second half of this paper,an integration approach is proposed to obtain bathymetric information in shallow water.This new approach uses two-medium photogrammetric techniques to obtain some reference points from stereo images,then a PCA(principal component analysis)based GWR(geographical weighted regression)inversion method is employed to model the shallow water surface model-DBM(digital bathymetric models).This approach can potentially make use of a large amount of visible-band only aerial images captured by UAV to obtain bathymetric information in an affordable way comparing to other bathymetric systems.Based on the above explanation,the research work is as follows:1.The significance of water depth and bottom topography information to the development of marine economy is analyzed.The research progress of camera calibration,multispectral remote sensing inversion of water depth and two-medium stereo photogrammetry at home and abroad is studied.The common methods to obtain water depth are introduced,and the research content and technical route of this paper are described.2.A constrained camera iterative calibration method is proposed which uses lines,planes,angles as constraints.This new method is applied to both land and underwater environments,respectively,and the results are analyzed comparing to other camera iterative calibration methods.In addition,the new method is used to calibrate wide-angle cameras to confirm that the new method has good flexibility and adaptability,and has practical applications in multiple environments.3.The basic principle and common refraction correction models of two-medium photogrammetry and the principle and common inversion models of multispectral remote sensing bathymetry method are introduced.Based on the analysis of the characteristics of projected images,this paper introduces the water depth refraction correction method based on projected images.Firstly,the water depth information is obtained by the water depth refraction correction method based on the projected images in the texture rich areas.The obtained water depth information is applied to the visible-band images to obtain bathymetric information,comparisons are made among different band combinations and band-ratio methods.The water depth refraction correction method based on the projected images is integrated with the visible-band based bathymetry method.The integrated approach offers a new and affordable way to obtain bathymetric information which only uses wide available visible-band aerial images,such images can easily be captured by any UAV systems.4.Several experiments are conducted to test the proposed constrained camera iterative calibration method and the integrated approach for bathymetry is.The first experiment is to calibrate a UAV camera using proposed constrained camera iterative calibration method.The image is resampled,and then the projected image is further generated.The water depth information of the swimming pool is obtained by the water depth refraction correction method based on the projected image,and the elevation of the water depth elevation point on the same depth line is corrected.The second experiment demonstrates the proposed integrated approach for bathymetry.In order to compare to different band combinations and other inversion methods,4-band(red,green,blue and near infrared)aerial images are used.Comparing to three conventional linear inversion models,and other band combination,the results indicate that the proposed integrated approach(water depth refraction correction method based on projected images and the visible-band based bathymetry method)is the best among all methods.This outcome also indicates the potentials of the proposed integrated approach in practical use.