Page 384 - NGTU_paper_withoutVideo
P. 384

New Geomatics Technologies and Applications











































                      Fig. 2.  Results of the proposed model and Douglas-Poker and Viswalingam methods for Zerivar Lake

               There are a number of metrics that can be used to assess the effectiveness of simplification processes. To evaluate the
          efficiency  of  the  proposed  model  and  commonly  used  simplification  methods,  four  indices  of  corrected  middle  Housdroff
          distances, area differences, similarity of angle changes, and similarity of average curvature were used in this study. The corrected
          Hasdroff distance index is based on the minimum distances of each point from all points of the line [25]. The minimum distance

          will be zero for the Douglas Poker and Viswalingam methods because they use the initial line points for fitting, while it will not
          be zero for the proposed model of this research because it does not use the points and fits them all. This index in the proposed
          method was very near to zero for Zerivar Lake. The evaluation results are presented in Fig. 3. The similarity percentage of the
          changes in angle and average curvature were 12 to 14% better than other methods. In terms of the area criterion, the proposed
          model has maintained the area better than the feature's original form. Also, in the Douglas Poker and Viswalingam methods,
          there is a significant difference between the area of the initial feature and the area of the generalized feature, which is due to
          these methods' focus on preserving the feature node and causes a lot of distortion in the area. The results of evaluating the outputs

          of the three methods, based on these three indicators, are shown in Figures 4 to 6 via radar chart.






                                                                                                               6
   379   380   381   382   383   384   385   386   387   388   389