Study On Automated Selection Of Linear Drainage Feature Based On Multi-objective Optimization By Genetic Algorithms

Map generalization is a kernel problem of cartography, which is also one of difficult problems in cartographic domain. Automated map generalization is a high intellectualized visual process, abundant visualize thought and inspiration thought exist in map generalization, at present, existing theory and methods can't solve the problems of automated map generalization completely. After analyzing and studying the interrelated literature in and abroad, the author pointed out that map generalization is a multi-objective optimization problem with constraint conditions essentially, A new approach to map generalization making use of genetic algorithms, supplying a intellectualized way to solve the automated map generalization problems, which can make up present generalization theory and methods' shortage to some extent.Aiming at elimination problems of linear river network mostly, this paper attempts to make use of genetic algorithm to solve partial problems in automated map generalization, develops some new algorithms with certain intelligence, and got ideal effects. Main contents of this paper includes:(1) A basic conception framework of automated map generalization based on genetic algorithm is designed.(2) A structural river network data model based on reaches is developed. The model considers the spatial relationship between reach and reach, reach and areal drainage features and reach's hierarchy relationship of river network adequately.(3) A structural river network data model based on automated generalization oriented is developed, which contains lots of information needed by river network generalization. A convenient and fast structural algorithm of river network is developed, which overcomes present structural river network algorithms' limitation which only aims at structural river and simple river network, refers to variety of complex problems appearing in river network's structure, and has a stronger practicability.(4) An automated river elimination model based on multi-objective optimization by genetic algorithms is developed. The model takes river's length, interval, importance and so on into account, and considers preservation of overall structural characteristics of river network in river selection, which has a higher automated and intellectualized level than present river elimination models, at the same time the model is extendable.(5) A man-made river network elimination model based on multi-objective optimization by genetic algorithms is developed. According to the purpose and principle of man-made river network generalization, on the basis of considering preservation of graphical characteristics of man-made river network before and after generalization, length of canal, density control of man-made river network and so on, A man-made river network elimination model based on multi-objective optimization by genetic algorithms is developed, combining the basic characteristics of genetic algorithms. The experiment result shows that the structured elimination of man-made river network is carried out by the model which makes use of capability of genetic algorithms efficiently. The model is suitable for man-made river network generalization, and gives several parameters influencing the generalization effects.