Research On Automatic Label Placement Based On Rules-Engine

Cartographic label placement has been reported since 1960. This old question is a difficult and hot research topic in cartography and related fields. The idea of Rules-Engine is to separate the business rules from specific program implementation code so that the rules can been completed independently. According to current specifications and experiences, the available map label rules only give fundamental requirements, such as easy-reading, no-covering, unambiguity etc. The descriptions of these rules are always blurry, abstract and arbitrary. However, in the view of technology, one label placement algorithm just implements following one group of specific rules. The specific code should be altered as the changes of labeling rules. This contradiction leads difficulties to take the dissimilarity between features into account. The emergence of the Rules-Engine provides chances to solve this problem. This paper employs the knowledge representation, discursion, and searching match techniques to complete labeling tasks, which will also be used to support cartographic generalization in digital environment.Label placement is an intelligent process, which determines the label’s style and layout base on importance and characteristics of the marked feature. The labeling style includes font, size, color, and shape. The labeling layout includes order, annotation direction, position, and interval. There are two kinds of text form for labeling. The first one is letter character, and the other one is symbolic character. These two methods are different. Chinese characters belong to symbolic character. Thus, automated label placement for Chinese characters is more complicated than characters such as English and French. In this paper, we take into account the feature’s attribute semantic and symbolic graphics such as size shape from the views of rationality, comprehensive, systematization and other factors. Aiming at our country’s map making, several label configuration modes including point configuration mode, line-point configuration mode, parallel line configuration mode, buffer line configuration mode, main skeleton line configuration mode, axis configuration mode, convex hull configuration mode and hash type configuration mode are summarized. Each mode provides the appropriate configuration method to estimate the style of the label express and candidate positions.Gestalt psychology is a cross fields of cognitive psychology and graphics, which focuses on the research of the impacts of the combination of graphics on human cognition. Base on Gestalt theory, this paper also proposed five label evaluating factors which influence the label placement. The measurement approach is considered in the proposed method as well. Finally, the label quality evaluation model was implemented and tested.Labeling configuration rules include the condition, priority, constraints, configuration method, and process strategies from the views of delivery of map information. These factors are utilized to achieve the label’s function, and beautify the overall results of the map. The first step of automatic label placement is to express these label configuration rules formally, which means to establish the rule base. In the paper, we also coupled the knowledge representations and discursion technology with the actual situations of label placement, detailed the configuration rule system, the structure of label configuration rule base, establishment, and maintenance methods to achieve the formally expression of the labeling rules in digital environment. The label knowledge base is composed of the configuration facts base, configuration rule base, conflicting rule base, conflicting handling rule base, and the configuration parameters base.The purpose of Rules-Engine is to separate the business logic from program logic through storing the business logic with rule files and resolve the complex business rules. Based on this purpose, this paper discussed the descriptions and formal representations of cartographer’s knowledge in a rule language RuleML on one hard; On the other hand, we also proposed a method to derive label configuration mode and parameters using the improved Rete algorithm. Finally, we established the automatic label placement framework based on Rules-Engine.The label optimization is also included in this paper. It is completed by considering the optimization strategy of overall features based on the cartographer’s experience. Moreover, we thought that all the labels should be divided into topology regions before optimization, which will overcome the limitation of separating the labels by layer.The proposed configuration model and algorithms were implemented and tested. The label knowledge base for 1:50 000 and 1:250 000 were built and an automatically labeling placement prototype system with NxBRE was developed in the end.