Discussion on the application of intelligent analysis in early warning of illegal buildings in land supervision

    The application of intelligent analysis in life has developed rapidly in the 20th century. Nowadays, intelligent analysis can be seen everywhere in people's lives. According to the category of intelligent analysis application, it can be divided into the following categories:

　　Cross-border analysis: usually installed on the walls of key areas of factories, schools, and detention centers to monitor people's cross-border behavior;

　　Intrusion analysis: generally used to provide early warning of intrusions in key protected areas such as museums and bank vaults;

　　Loss analysis: used in museums, exhibition halls, etc. to provide early warning when important items leave the designated area;

　　Direction analysis: monitor and manage the flow of people or vehicles at important entrances and exits or one-way roads;

　　Detention analysis: Analysis of people and objects staying in important monitoring areas and related prohibited areas;

　　Intelligent tracking: Applied to target and track people and objects in residential areas or related public places for alarm.

　　State-owned land covers a wide range. my country has a huge area of ​​state-owned land, and its distribution is not concentrated. Therefore, the management of state-owned land is difficult to supervise and a series of problems such as delayed discovery of illegal occupation. Among the illegal occupation of state-owned land, illegal construction on state-owned land is the first target of land supervision. Many developers will take advantage of the lack of timely supervision and illegally occupy state-owned land on suburban state-owned land.

　　Prevent problems before they happen and provide timely warnings

　　There are relatively obvious characteristics of illegal occupation and construction on state-owned land. For example, in the early stages of illegal construction, some people will conduct site scouting, followed by the involvement of engineering vehicles and activities such as laying the foundation. At this time, there will be characteristics of surface changes, and characteristic behaviors such as green protective nets and mobile houses will also appear during the entire construction process.

　　Our goal is to give early warnings at the early stages of illegal occupation, which is of great significance for timely discovery and timely investigation. If the illegal facts have already been created, the future investigation and punishment will be more difficult. Intelligent analysis of the above features by algorithms is of great significance to land supervision of illegal buildings.

　　Different from day to day,

　　When illegal occupation of state-owned land occurs, the scene at the scene will generally change significantly, and can be divided into two categories according to the length of time accumulated.

　　Movement changes

　　Refers to significant changes in the monitoring scene within a short period of time, such as three or five workers working in the monitoring scene, or construction vehicles operating in the area.

    Surface changes

　　It refers to the fact that after a long period of time, the ground surface has undergone large and obvious changes due to activities such as foundation laying or excavation.

　　Motion detection has extremely high detection accuracy for timely detection of manual site inspections and monitoring of construction vehicle transportation after construction sites are built. By means of denoising and background modeling of the monitoring scene, the computer can learn the model of the monitoring area. When "foreign objects" (people or construction vehicles) appear in the monitoring scene, background subtraction, morphology and other operations can be used to accurately obtain the moving objects in the monitoring scene.

　　Surface changes can sensitively detect significant changes in the surface, and have a good detection effect on surface changes caused by the foundation laying stage. By learning the monitoring scene over a long period of time, it can determine whether there are surface changes in the scene. Surface changes are an important sign of the beginning of illegal occupation of state-owned land. The study of surface changes through intelligent analysis is of pioneering significance.

　　Eliminate the coarse and retain the fine, eliminate the false and retain the true

　　Through literature research and field visits, we found that construction vehicles must be involved in the construction process, and there may be features such as mobile houses and protective nets. The identification of construction vehicles, after motion detection in the video surveillance area, can be well identified by using the HOG feature plus SVM classification method. Generally speaking, construction vehicles represented by excavators usually appear in the initial stage of illegal construction of foundation digging. By grasping the recognition of the excavator's mechanical arm and bucket, this feature can be accurately captured. Construction vehicles represented by earthmoving trucks and mixers run through the entire construction process.

　　According to relevant regulations, when the building height exceeds 4m, a protective net must be set up that gradually rises with the wall to prevent personal injury and object falling, reduce noise and dust pollution, and achieve the effect of civilized construction. Generally, the protective net body is mainly green, accompanied by horizontal, vertical and oblique cross supports. Characterized by color and stripe texture, the classifier is designed with slide-window+FFT+adaboost structure, which has a good detection and recognition effect on the protective net.

　　Detection of prefabricated houses. In modern buildings, prefabricated houses are widely used by construction units as temporary residences for personnel because of their advantages such as lightness and quick installation. Therefore, areas where prefabricated houses appear are likely to be under construction. The structure of prefabricated houses is relatively regular, and the colors are single and orderly. SIFT can be used to stably describe the characteristics of prefabricated houses.

　　Challenges in Application

　　Due to the complexity of actual scenes and the performance limitations of imaging devices, the videos captured by cameras often have a large degree of jitter. Coupled with factors such as the variety of shooting angles, different lighting conditions, object occlusion, and interference from a large number of non-interesting targets, the quality of the images obtained varies greatly. Many algorithms that perform well in theoretical analysis do not perform well in practical applications. To solve these problems, the system must be cleverly designed and precisely arranged.

　　Conclusion

　　Although the application of intelligent analysis in land supervision and illegal construction early warning is subject to various external objective conditions, and its accuracy is also reduced, its application is believed to promote technology. We might as well imagine here: in the near future, in the application of land supervision and illegal construction early warning, things that cannot be perfectly achieved by humans, intelligent analysis can help us do it!