Condition Numbers for Evaluation of Ill-Posed Problems in Photogrammetry

We discuss the principle of condition numbers that used for evaluating the extent of ill-posed problem of normal matrix.There is a contradiction between the stability of solution and the condition number of resection in photogrammetry.We find that it is not suitable in all cases to evaluate the extent of ill-posed problem by condition numbers.Three types of possible risks for evaluation of ill-condition extent with condition numbers were addressed in detail.Removing of outliers and re-parameterization are the prerequisites for evaluation of ill-condition extent with condition numbers.There are two effects of re-parameterization for ill-posed problems.One is improving the problem of ill-condition caused by numerical computation,and the other is avoiding the risk of using norm to evaluate the extent of ill-condition.Results of simulated experiments show that the proposed approach is validate for improving the problem of ill-condition.