Research On Intelligent Retrieval Of Engineering Drawings Based On Visual Content

With the development of mechanical manufacturing technology,a large number of engineering drawings have accumulated in various fields such as transportation,manufacturing,aerospace and so on.Most of these engineering drawings have not established a database retrieval mechanism for resource reuse and data query.Statistics show that more than 80% of the design activities in the development of new engineering drawings are reuse or minor modification of existing designs.Reasonable reuse of existing resources can effectively shorten product development cycle,reduce product cost and improve product quality.Therefore,it is necessary to establish database mechanism for engineering drawings and design a similar drawing retrieval algorithm.At present,most of engineering drawing retrieval systems are based on key words,which are difficult to select key words and can not describe drawing shape information.Based on the characteristics of engineering drawings and the visual information of engineering drawings,this paper designs an efficient and accurate engineering drawing retrieval system.Firstly,based on the combination of the key text information and visual content information of engineering drawings,this paper establishes the engineering drawing database,and then realizes the multi granularity intelligent retrieval algorithm based on the visual content of engineering drawings to realize the reuse of engineering drawings.Combining theory and experiment,this paper proves that the algorithm is better than traditional ones.The main contents and innovations of this paper are as follows:(1)This paper introduces the content-based image retrieval and engineering drawing related theoretical system and key technologies,including key text information extraction,feature extraction,similarity measurement,etc.(2)This paper proposes a new multi-stage algorithm for extracting key text information from engineering drawings.Aiming at the problem of regional distribution and small proportion of key text information in engineering drawings,this paper designs a multi-stage key text information extraction algorithm,and carries out the comparison and analysis of relevant experiments.It is proved that the multi-stage key text information extraction algorithm can extract the text information of engineering drawings more accurately.(3)In this paper,a new content-based feature extraction algorithm is proposed.In this study,the algorithm of drawing retrieval based on view segmentation is designed to solve the problem of non-uniform layout of view.(4)This paper designs a new similarity measurement algorithm for engineering drawings.Different from the traditional similarity measurement algorithm,this paper proposes a similarity measurement algorithm with optimal weight.It can solve the problem of inconsistent information contained in views.(5)This paper realizes the fast retrieval of large-scale database.Multi granularity retrieval of view classes and instances can greatly reduce the scope of retrieval and improve the speed of retrieval.The experimental results show that the intelligent retrieval system based on visual content is a fast retrieval scheme,80 times faster than the traditional retrieval algorithm,and verify that the algorithm based on visual features is suitable for large-scale engineering drawing retrieval.There are 33 pictures,8 tables and 64 references.