Researching The Method Of Gross Error Detection In The Data Processing Of Oblique Aerial Image

In recent years there has been a significant development of oblique aerialphotogrammetry in the field of international surveying and mapping, the high-tech ismainly used for obtaining oblique images which can provide abundant textureinformation for3D model reconstruction. In the data processing of oblique aerial image,mass match points are collected automatically by the way of multi-view image match.But owning to the characteristics of oblique aerial images, such as photography scaleinconsistency, resolution difference and serious land feature block, the points datacontains more gross errors which will affect the accuracy of the subsequent automaticaerial triangulation seriously. Therefore, the detection of multidimensional gross errorsin mass data is vitally important for oblique aerial photogrammetry.On the one hand, almost all the current algorithms of gross error detection areunsuited for detecting multidimensional gross errors in mass data due to the detectioncapability or processing efficiency limit; on the other hand, several types of attitudeparameter result in the relative positional relationship being more complex between theoblique aerial images, so that conventional the model of continuous relative orientationis no longer applicable. In response to these problems, this paper is trying to solve it byresearching both the algorithm of gross error detection and the function model ofadjustment.1)In this paper, based on the gross error detection principle of correlative analysis,the method of simultaneous locating and evaluating multidimensional gross errors inphotogrammetry is proposed which is referred to as LEGEP. Through the experiment ofsimulation gross errors shows that LEGEP can accurately locate the multidimensionalgross errors in mass data and calculate the numerical size of gross errors at the sametime; In the other experiment, comparing LEGEP with several other representativealgorithms of gross error detection, it is found that more gross errors had been detectedwith less iterative calculation by using LEGEP and the accuracy of adjustment isimproved obviously. So it proves the superiority of LEGEP in both detection capabilityand efficiency. 2)The method of using direct solution relative orientation model as the functionmodel of adjustment is proposed in this paper. The model has no need to be given theapproximation of any parameters. It is a universal adjustment model which is suitablefor detecting gross errors in the data processing of oblique aerial image.In the final experiments of the paper, it proves that based on the direct solutionrelative orientation model, application of LEGEP to the mass data in oblique aerialphotogrammetry is an effective method for multidimensional gross errors detection.This could have more practical applications in the data processing of oblique aerialphotogrammetry.