A Study On The Automatic Identiciation Algorithm On Fixed-Point Monitoring Of Urban Illegal Buildings

The strictest land management policy is being executed to ensure that land utilization is under the supervision of the government.Therefore,it is significant for our country to detect and investigate the illegal land occupation as soon as possible for the national food safety and maintaining social stability.Considering the reality of low intelligent level of illegal building monitoring,this paper is devoted to the study of the key technologies of fixed-point monitoring of old and new phase contradistinction to discovery urban illegal buildings automatically.The main contributions are as follows:(1)The method of change detection is studied by intergrating Hu moments and similarity measurement to mitigate illumination interference.By analyzing the anisotropic characteristics of texture features of illegal buildings in different phrases,shape changes are used to mitigate the interference and extract the texture information of illegal buildings.The change threshold of jointed OTSU is determined.And the change detection is established by intergrating the Hu moments and the standardized difference square sum matching to perform the difference evaluation of texture information.Experiments show that the change detection is highly robust which can effectively reduce the influence of environment light,and the accuracy is 95.2%.(2)An illegal building recognition algorithm based on deep learning and multiple features is researched.By using contour topological features,color information,the spatial restraint relationship,and the distribution characteristics of the coordinates of vanishing points caused by the relationship,eliminating common interference factors by vehicle detection in a deep learning framework,experiments are designed to evaluate the algorithm and test results show that the recall for identifying multi-type illegal buildings reach 98.5%,the precison is 97.5%.(3)A method of urban illegal building positioning algorithm based on preset points placement and monocular ranging is proposed.Considering the difficulty of locating illegal buildings automatically after finding suspicious land use by fixed-point CCD,pre-set points placement within monitoring range is studied.The high-resolution optimal placement of preset points of fixed-point CCD is achieved.Meanwhile,by using road depth information of the results of illegal building identification,illegal buildings can be positioned automatically within 20 meters error range.The designed experiment verifies the effectiveness of the algorithm.Based on above algorithms,a set of automatic discovery of illegal buildings monitor system in Anshun City is designed by using C++ in the Windows system.Experiments are designed and the result verifies the effectiveness of the algorithms.The precision of the system is better than manual detection and it is every day tool of illegal building monitoring now.