煤矸石山自燃灾害严重破坏生态环境，影响安全生产，已成为我国生态文明建设进程中的瓶颈问题，为科学制定灭火方法及实施灭火工程，自燃煤矸石山火区范围的准确探测至关重要。

提出采用无人机热红外倾斜摄影测量技术探测高温区三维分布的思路，研发热红外图像多尺度高斯滤波细节增强技术，突出目标边缘信息；基于FLIR Tools 软件开发工具包SDK (software development kit)开发热红外图像一键式批量化处理软件，利用ThreadPool高效地开展热红外图像的多线程异步处理，全局统一所有热红外图像颜色阈值与温度数值的对应关系；最后采用批量处理后的热红外图像进行自燃煤矸石山火区表面温度场三维重构，并在陕西某煤矸石山火区进行了现场应用。

结果表明：(1) 经过多尺度高斯滤波细节增强后，热红外彩色图像细节特征得到显著增强，热红外图像信息熵由6.989提高到了7.624。(2) 所开发的热红外图像批量化处理软件可在10 min内对1 000张热红外图像进行快速处理，极大提高了工作效率。(3) 上述技术在陕西某煤矸石山火区现场成功构建了一区及二区煤矸石山火区三维表面温度场，分别圈定了11 693 m²和10 800 m²高温区，为火区发展态势评估和后续灭火方案设计提供了支撑。研究成果可应用于煤矸石山火灾、煤田火灾、垃圾填埋场火灾、森林火灾以及建筑火灾等高温场景的温度场三维重构，为火区侦查、火情调查及火源位置诊断提供详实的火场数据支持。

The spontaneous combustion of coal gangue dumps has severely damaged the ecological environment and affected the production safety, which has become a bottleneck in the process of ecological civilization construction in China. In order to formulate a fire extinguishing plan and implement a fire extinguishing project, it is very important to accurately detect the fire area of the spontaneous combustion coal gangue dump.

Herein, the idea of detecting the three-dimensional distribution of high-temperature areas using unmanned aerial vehicle (UAV) based thermal infrared oblique photogrammetry technology was proposed. Meanwhile, the multi-scale Gaussian filtering detail enhancement technology for thermal infrared image was developed to highlight the target edge information. Besides, the one-click batch processing software for thermal infrared images was created based on the FLIR Tools Software development kit(SDK), and the multithread asynchronous processing of thermal infrared images was efficiently carried out using ThreadPool, thereby realizing the global unification of the corresponding relationship between the color and temperature value of all thermal infrared images. Finally, the surface temperature field of fire area of the spontaneous combustion coal gangue dump was three-dimensionally reconstructed based on a batch of processed thermal infrared images. This method was also applied on the site of a burning coal gangue dump in Shaanxi Province.

The key findings indicate that: (1) Through multi-scale Gaussian filtering detail enhancement, the detail features of the colorful thermal infrared images are significantly enhanced, and the information entropy of thermal infrared images increases from 6.989 to 7.624. (2) The developed batch processing software of thermal infrared image could quickly process 1 000 thermal infrared images within 10 minutes, greatly improving the work efficiency. (3) The above technology has been applied to successfully construct the three-dimensional surface temperature fields for two zones in a coal gangue dump fire area in Shaanxi, delineating the high-temperature areas of 11 693 m² and 10 800 m², respectively, which provides support to the assessment of the fire development trend and the creation of subsequent fire extinguishing plans. The research results could be directly applied to the three-dimensional reconstruction of temperature fields in high-temperature scenarios such as coal gangue dump fires, coal fires, landfill fires, forest fires, and building fires, providing comprehensive fire data support for fire investigation and fire source locating.