1985 Vol. 10, No. 3
The automation of photogrammetry can be realized thorough digital way.However,the speed of processing a vast amount of data and the reliability of image correlation are the key problems awaiting solution.The image digitizing mapping system proposed in this paper includes the following distinguishing features.(1)For the orientation of digitized images and the arrangement of sampled pixels along epipolar lines,data from aerial triangulation are effectively used thus,the computation load has been greatly reduced,(2)Considering both reliability and aceuracy in the digital correlation,an appropriate twodimensional target segment is suggested.(3)In order to get more reliable image matching,a new method,so called mufti-criterions method,is used in place of the generally adopted single criterion one.(4)The gross errors are detected by parallax signal processing.(5)Chinese character notes can be printed on orthophoto maps during digital rectification.
Starting from the viewpoint of Professor Wang Zhizhuo:"Both photogrmmetry and remote sensing are belonging to a same entity,which should not be segmented from the academic standpoint",the close relationship between photogrammetry and remote sensing as well as thier service object and developing orientation are discussed in this paper,based on the experiences of remote sensing research works carried out recently in the Wuhan Technical University of Surveying and Mapping.
In this paper,the process of carrying out a DEM international experiment,organized by the ISPRS Commission Ⅲ in 1980 is described.The instrument used is a home-made Computer Controlled Digitel Mapping System including an Image Coordinate Digitizer.It has been found that so far as precision and speed is concerned,our hme-made instrument is superior as compared with other imported instrumnets.The cost is also much lower.
Mit der Theorie der Trennbarkeit von Modellfehlern im Gauβ-Markoff-Modell wurde die Trennbarkeit von groben Fehlern und systematischen Bildfehlern bei der Bündelblockausgleichung untersucht.Die Abhängigkeit dieser Trennbarkeit von Paβpunkt-und Bildfluganordnung sowie von der Blockgröβe wurde diskutiert und daraus wurden einige Möglichkeiten zur Verbesserung der Trennbarkeit gezogen.Diese Untersuchung ist für die hochgenane photogrammetrische Punktbestimmung von groβer Bedeutung.
In this paper,the gross errors detection of a stereopair relative Orientation with the use of Simplex Algorithm of Linear Programming(i.e.∑|vi|=Min method) is described.The result of thir experiment bas shown that this method is better than either the Data-Snooping method or the Danish method in determining the portion of gross errors when the number of gross errors is larger than one.The computer time needed for calculating the elements of relative orientation of each stereo model is the same as cmpared with the Danish method.
Based upon the correlation analysis of coefficient matrix of a space resection problem,a new space resection method and a methematical model are given.The posteriori weights of observations are then estimated to improve the accuracy of adjusted results.Finally,the correctness of this method and model is verified through data processing with a computer.
In this program system,the interior geometrical relationship of central projection and the abundance of observations are adequately ultilized so that its ability to detect primary data errors has been enhanced.Based on the fundamental principle of stereo-model coordinate transformation,unified formulas are used to transform the photogrammetric coordinate systems,to join the airstrips and to form the error equations of coordinate transformation of the independent models.Based on the principle of block adjustment by the use of independent models,the feature of band sparse symmetric positive definite matrix in solving large normal equations and the advantages of FORTRAN language,the lines of the computer program and the times of iteration for repeated adjustments have been greatly reduced in this program system.This program system is divided into 6 blocks.Recently it has been performed successfully on various computers such as Z-80,PDP 11/23,PERKIN ELMER OS/32,M一150,VICTOR 9000,VAX-11/750,68000 etc.
For three-dimensional block adjustment of independent models there are seven unknown parameters in every model.To solve these unknowns,the algorithm socalled alternating "plan" and "height" adjustment has popularly been used.Based on the analysis of the characteristics of gravity-centralized coordinates used in the adjustment prcoedure,a new algorithm is proposed in this paper,with this method the seven parameters are divided into two parts:three for shift(dxg,dyg,dzg) and four for rotation and scale change(dΦ、dΩ、dκ、dΔλ).It has been verified theoretically,that the three shift-correction values can be obtained by calculating the mean values of the coordinates of all tie points at every iteration and needn't be involved into the normal equations.Therefore,only four unknowns for each model remain to be solved.Two programs have thus been written based on the conventional and the new algorithms and run with synthetic data for comparison.The result shows the correctness and advantages of the new algorithm.The new method takes not only much shorter computation time but also much smaller core storage units as compared with the conventional one.