With new directives from Swedish authorities imposing municipalities to digitize sections of their plan archives, the question of digital detailed plans is becoming more relevant than ever in Sweden. Digitizing already existing detailed plans is time consuming, so effective automated digitizing methods will be valuable to save time in this process. However, in order to know if a method is effective it first has to be evaluated. This study aims at evaluating a recently introduced method for automated digitizing of detailed plans, and it is the first one evaluating this method in a quantitative manner. The questions to be answered within the study is whether the implemented method is effective and if it has any weaknesses. Additionally, whether a number of defining characteristics of the detailed plan maps influence the quality of the result. As the quality of digitized detailed plans have not been subjected to systematic evaluation before, a novel contribution of this study is also suggesting a framework for how this can be measured and evaluated.
The method consists of 3 steps and the first 2 steps, namely automated georeferencing and automated vectorization, have been performed on a set of 75 detailed plans. Using manually digitized versions of the same detailed plans as ground truth, the results of these two steps have been compared and evaluated using a set of quantitative measures.
Findings from this study have shown that about 70% of the detailed plans tested can be georeferenced, and 44% of relevant areas in the plan maps can be vectorized using the method. However, the results have displayed a significant disparity of quality, with error values for georeferencing ranging between under 5 meter for the best results and over 100 meters for the worst.
The weaknesses that have been identified for the method are mainly that the georeferencing procedure requires extensive manual supervision, that the vectorization produces polygons of ambiguous belonging, and that the method is limited to multicolor detailed plans. Furthermore, a small plan area has been identified as the most influential factor for a low quality result. Main conclusions of this study have been that the method can be considered effective for digitizing detailed plans to some extent. Additionally, the method for evaluating the quality of digitizing could be expanded by reviewing more factors such as shape and gaps between polygons in future work.