Warning method based on specific illegal building identification in land resource video monitoring system

Abstract

The present invention provides a warning method based on specific illegal building identification in a land resource video monitoring system. The method comprises: by setting up cameras in field monitoring points, capturing pictures periodically at fixed points, and returning the pictures to an information center server. Based on the traditional land resource monitoring network, the method provided by the present invention uses an R-CNN (Fast RegionConvolutional Neural Networks) framework based on the convolution neural network to carry out target detection on the building in the information center server; identification results in the pictures within a continuous period of time are compared and analyzed to determine whether the building is an illegal building, and abnormal results are submitted to carry out manual audit so as to achieve the purpose of intelligent and automated monitoring of land resources; and according to the method provided by the present invention, de-atomization treatment based on a dark channel is carried out on the returned captured pictures so as to reduce the influence of the haze weather on the detection result.

Description

technical field [0001] The invention relates to the field of computer application technology, in particular to a method for identifying and early warning specific illegal buildings applied in a video monitoring system for land and resources. technical background [0002] As an industry application of the existing mature network video surveillance technology, the video surveillance system of Land and Resources generally includes front-end video information collection, video encoding network transmission, information center control management and data processing. The system is as attached figure 1 shown. In actual project construction, with the rapid increase in the number of monitoring points, manual on-duty watching will be faced with a large amount of video or snapshot image information, and supervisors will inevitably neglect and omit important information when faced with a large number of monitoring images, and visual fatigue will also It will make the regulatory effect ...

AI Technical Summary

Problems solved by technology

However, the same type of objects in the real environment usually also show complex shapes, so a large and accurate training set must be used to achieve good recognition results

Images