Research On Spatial Retrieval Method Based On Hand-drawn Graplhics

With the development of computer technology,multimedia technology and communication technology,people are generating and sharing information every moment in their daily lives.The most extensive information is geographic information.Geographic information is closely related to people's clothing,food,housing and transportation,and has become an indispensable part.The geographic elements in the geospatial database are such a very important geographic information.At the same time,with the rapid economic development,the continuous expansion of geospatial databases and the popularization of high-tech hardware such as touch screen mobile phones and tablet computers,the technology of searching by means of searching for maps has gradually entered people's lives.People's demand for high-convenience life services also makes the current situation of retrieval technology of geographic elements in geospatial databases particularly important.As one of the important branches of content-based image retrieval technology,hand-drawn graphic retrieval technology currently belongs to the preliminary research stage.The shape feature of hand-drawn graphics is the most intuitive perception of people's spatial information.Therefore,the text selects the shape feature as the feature descriptor to retrieve the hand-drawn graphics and the geographic feature set in the geospatial database.The main work of this article is as follows:(1)First of all,in order to solve the problem of wrong strokes that users will encounter when drawing hand-drawn graphics,this paper pre-processes the hand-drawn graphics drawn by users,including the elimination of convergence points,closed correction processing and smoothing processing,using a comparison of relative density and threshold size The method eliminates the gathering point;adopts the method of accumulating the offset angle and extending the stroke in reverse to perform the correction process of closing;the discrete point adopts the method of equal interval to smooth the hand-drawn graphics.Removal of noise data in strokes improves the accuracy of subsequent graphic retrieval.(2)The background and current situation of shape-based retrieval technology are combed and summarized,and then the shape feature description methods are elaborated and compared and analyzed from the perspective of regions and contours.The advantages and disadvantages of different methods are pointed out.The region-based shape description method is selected.The vector graphics invariant moment method and the curvature scale space method based on the contour-based shape description method are used for space retrieval.Then,the similarity measurement methods of different feature description methods are selected and elaborated.(3)In this paper,in order to better extract the information contained in the hand-drawn graphics,and to solve the problems of large data volume,information redundancy and retrieval stability,a region-based shape description method and contour-based shape description are proposed The combination of methods,that is,the method of vector graphics invariant moment and curvature scale space method,forms a retrieval strategy from coarse to fine.After preprocessing the hand-drawn graphics,under the premise of determining the retrieval threshold through multiple experiments,vector graphics invariant moment retrieval algorithm and curvature scale space are used for the Wuhan University housing dataset,the national lake dataset and the Wuhan University road dataset,respectively The retrieval algorithm and the retrieval algorithm combining vector graphics invariant moment and curvature scale space were compared and analyzed experimentally.Three indexes,namely precision,recall and F1,were selected for evaluation.The experimental results on different data sets are higher than the evaluation when the two methods are retrieved separately.The overall retrieval effect is better,so the effectiveness of the algorithm is verified.