Abstract: With the emergence of digital urban design methods, the case-based reasoning (CBR) strategy has broadened research ideas for the design of urban forms, as well as provided inspiration for the deep mining and utilization of urban data. The CBR strategy uses existing knowledge to solve problems, which makes it possible for urban morphological research to lead from the analysis of "index" to the generation of "form". With the establishment of morphological case database, the calculation of quantitative indicators, together with case similarity retrieval, makes building a framework for morphological case matching and generative design based on CBR a feasible research direction. Taking block scale as an example, the current study selected six representative cities in OpenStreetMap as data sources, extracted block unit data and its internal building data with the help of geometric operation, and then stored them on the database. For each block case, three quantitative indicators were calculated and inserted into the database as attributes: the urban form control index, block geometric shape descriptor, and block primary and secondary axis. The urban form control index was based on the collection and calculation of the basic geometric information of the case, which consisted of three data: area, GSI and FSI. The block geometric shape descriptor was based on the geometric features of the block contour (e.g., area, perimeter, convex hull and bounding box). Eight descriptors were calculated and integrated into the shape feature vectors for each case, which were used as comparison factors for case similarity. Calculation of the block primary and secondary axes makes it possible to record the direction of blocks in advance, which is helpful towards directly obtaining the matching angle in the process of matching and generating. By comparison, the OBB method was chosen as the calculation method, and the exhaustive calculation was combined in order to conduct the direction matching. Based on the case collection and indicator calculation outlined above, a five-step workflow for single case matching generation can be constructed: target block preprocessing, form control index screening, shape similarity retrieval, optimal direction matching, and building volume generation. An experiment was conducted in the current study in which two European cities were selected as test sites and eight blank target blocks with different scales were selected for the experiment. Based on the abovementioned workflow, the program code screens and sorts the top eight most similar cases from the database according to the input conditions and morphological characteristics of each target block. One case was manually selected based on the site situation as the best matching case, and the model was then generated by analyzing the geometric information of the building to obtain the final generation result. The results revealed the different effects of the three quantitative indicators. As a pre-screening condition, the urban form control index narrowed the optional range of the search results. The block geometric shape descriptor provided an important basis for similarity retrieval, and the retrieval results varied according to morphological particularity and case richness. The block primary and secondary axes played a role in the process of direction matching, which can improve the efficiency of automatic direction searching. The research method used in this study presents a feasible technical strategy for the aspects of quantitative urban-form data mining, the application of database system, and case-retrieval-based generative design. This strategy provides a digital perspective for the transformation of index and form in the design of urban forms and provides support for decision-making around urban design. 

Key words: urban morphology, case-based reasoning, generative design, index and form, block scale

摘要： 在数字化城市设计方法涌现的当下，基于案例推理的技术策略为城市形态设计拓宽了研究思路，也为城市数据的深度挖掘与利用提供了灵感。研究建立了街区尺度的形态案例库，并通过形态量化指标的计算与案例相似性检索，搭建起基于案例推理方法的形态案例匹配与生成设计框架。探索了数字技术背景下城市形态量与形的转化与生成策略，从而为数据驱动的城市设计决策提供帮助。

关键词: 城市形态, 案例推理, 生成设计, 量与形, 街区尺度