*3.4. Similarity Analysis and Case Retrieval*

The data of HDFV have the same contribution weights for the similarity analysis, as they reflect the overall morphological characteristics of each sample rather than the independent indicators. Therefore, the Euclidean distance was used to calculate the difference between the input images. The closer the distance was, the more similar the two cases were. The case retrieval system can rank one or more cases for the target based on the distance. We constructed a case retrieval system that realized case pairs according to the plot shape or plots with buildings. When integrated with other sample attributes, architects can choose the proper ones among the recommended cases according to the specific task.
