Comparison of block adjustment methods and accuracies of photogrammetric point determination
This paper compares accuracy results of photogrammetric block adjustments using independent models, and bundles without added parameters, with block-invariant added parameters and with image-wise added parameters. Three real test blocks at scales of 1:4000,1:8000 and 1:60 000 were adjusted individually using the aforementioned methods. The root-mean-square (RMS) errors of discrepancies at checkpoints were compared and analyzed. It appeared that bundle adjustments with block-invariant added parameters gave the best results which were then used as the basis for comparing the RMS values of standard errors of independent model and bundle adjustments and of confidence ellipsoids of bundle adjustment, and accuracies estimated using RMS errors of residuals of independent model adjustments. The RMS values of standard errors of bundle adjustments appeared to be consistent, except for small-scale photography (smaller than 1:10 000). On the other hand, accuracies estimated using RMS errors of residuals of independent model adjustments appeared to be consistent, except in planimetry for large-scale photography (larger than 1:10 000). Hence, in the absence of checkpoints, standard errors of bundle adjustments may be used to estimate accuracy of photogrammetric point determination, except for small-scale photography which may then be estimated using the RMS errors of residuals of independent model adjustments.