Research on investigative and evidentiary methods for ecological environment and resource protection public interest litigation based on remote sensing images
As one of the important ways to protecting ecological environment and natural resources,civil public interest litigation in the field of ecological environment and resource protection has attracted significant attention in recent years.Due to the immaturity of public awareness and participation,the construction of public interest damage appraisal institutions,and the support of relevant administrative departments,procuratorial public interest litigation faces challenges such as passive lead sources and difficulties in investigating and collecting evidence.In the past two years,procuratorial organs have actively explored the use of satellite remote sensing,drone aerial photography,and other ways to address the issues in public interest litigation investigation and evidence collection.However,currently,most information is obtained through manual or human-computer interaction,which is time-consuming and laborious.Addressing the aforementioned challenges in public interest litigation cases related to the ecological environment and resource protection,this paper proposes a method for public interest litigation based on remote sensing images.This method is mainly composed of two parts.Firstly,a change detection model based on U-Net++ is used to generate the change detection pattern of the research area.Subsquently,the final clues for public interest litigation are derived through the analysis of change detection pattern.Specifically,a small number of samples are first enhanced through geometric transformation methods such as rotation and random addition of noise,constructing a training sample dataset with large capacity,strong diversity,and high complexity.Using the U-Net++ model for change detection,combined with the relevant factual requirements of public interest litigation in the fields of ecological environment and resource protection,such as the current status and changes of target damage,the change detection patterns are screened to serve as clues or a basis for the investigation of public interest litigation cases.Comparative experiments were conducted with existing fully convolutional network models and U-Net models.The results show that the U-Net++ model performs better than the other two models in terms of accuracy,recall,and F1 score.The optimal model trained based on the U-Net++ model achieves a detection accuracy of 83.08%and an accuracy of 93.91%.In the end,a satellite remote sensing and unmanned aerial vehicle linkage-assisted forensics platform was constructed,generating auxiliary forensics reports based on the platform as clues or a basis for investigating public interest litigation cases.The research results have been applied in the procuratorial organs of Guizhou Province,indicating that this method can provide data and technical support for the procuratorial organs to quickly discover clues and uncover the truth.The experiment shows that the method in this paper is feasible and can provide robust technical support for the discovery and investigation of public interest litigation clues.The multi-temporal remote sensing image change detection based on deep learning proposed in this paper can be applied not only to assist public interest litigation but also in other field such as natural resources audit and the ecological environment.
NETL
NSTL
万方数据
