Study On The Self-calibration Technology In Photogrammetry

In photogrammetry and remote sensing technology application, the camera geometry calibration is an important step to eliminate the systematic error and improve the precision of photogrammetric system. The self-calibration technique is now recognized as the most flexible and effective method, establish a correct self-calibration model is the core of high precision photogrammetry.At present, many additional parameters of self-calibration have no obvious physical or mathematical sense, excessive parameterized problem can easily arise, and tend to have strong correlation between these parameters. The strong correlation between the parameters of the model is also exist in the Standard Brown self-calibration model, which can bring adverse effects to the end result.The topic of this thesis is the self-calibration technique of digital camera. Base on a number of related literatures, Focusing on analyzing the mathematical theory of self-calibration technique, seek the theoretical foundation of the self-calibration technique, solve the problem of strong correlation between the parameters of the self-calibration model and the excessive parameterized problem. Draw a few useful conclusions from experimental tests. The main work and innovation of this paper is as follows:1. Based on the theoretical analysis, a conclusion has been drawn: The essence of the photogrammetry self-calibration model building is to functional approximation. This is also the theoretical basis of all work done in this paper. Also, the conclusion provides a solid mathematical foundation for many classical self-calibration model(e.g. Ebner model, Grun model) in the use of digital photogrammetry era. Although these two theories proposed a long time ago, but because they are in line with function approximation theory, it still has vitality.2. Based on the theory of function approximation, constructed Legendre self-calibration models with Legendre orthogonal polynomial. Through experimental verification, Legendre self-calibration model has better applicability than Ebner model and Grun model, when dealing with complex distortion, can provide higher precision calibration.3. Based on the theory of function approximation, constructed Fourier self-calibration models with Fourier functions. The additional parameters in Fourier self-calibration models is completely independent, so it is more acuration than Legendre self-calibration models. Through experimental verification, to achieve the same calibration accuracy, additional parameters used less in Fourier model, then can effectively avoid over-parameterization problem.4. An improved in-plane distortion model is proposed. The improved model can effectively suppression the strong correlation of the Brown model, to achieve better calibration results, which is more suitable for close range photography camera calibration.