污水暗排口是一种较难监测的水污染源。为更好地探索监测这种污染源的方法，本研究尝试利用热成像技术探测污水暗排口。其原理是根据污水与正常水体温度的不同，借助红外热像仪（仪器型号为IR SnapShotTM）获取的水体影像所反映的温度差异，结合调查资料，确定污水暗排口。本次研究进行了两组实验：对照实验和排污模拟实验。在排污模拟实验中，利用热成像技术对污水暗排口排污过程中水面温度分布进行了连续监测，发现污水与净水的温度分布存在显著差异，净水温度分布均匀，而污水排入后水面温度呈辐射状同心圆分布，从温度影像上可以明确地把两种水体区分开来。在此基础上，还对排入水体的污水面积进行了解析。计算污水面积的关键是确定区分净水与污水的温度阈值，文中探讨了两种确定阈值的方法：人工法和差值法。分析发现，虽然两种方法确定的污水面积有一些差异，但是均能比较客观地得出污水的面积。人工法虽然灵活，但主观性较大；差值法较客观，也较准确。在实际应用时可根据实际情况选择合适的方法。

Underwater sewage outfall, a kind of water pollution sources, is difficult to monitor. To find a piratical method to monitor this kind of pollution sources, this study focused on applying thermal infrared imaging technology to identify the underwater sewage outfall. The theory of the adopted method is that the temperature of the waste water usually higher than the normal water so that their characters are different in the thermal image. In this paper, two kinds of the experiments were performed: one was for contrast and another was the simulation experiments in field. In the simulation experiments, the surface temperature of the underwater sewage outfall were scanned using the IR SnapShotTM, then the temperature information was extracted from the thermal images and analyzed. The results showed that there was a great difference between the distribution characters of the temperature of clean water and polluted water: temperature distribution of clean water was characterized as smooth, while that of polluted water was radial. Thus it could distinguish the two kind of water bodies by the thermal infrared image. Based on the conclusion above, the areas of the polluted water were calculated and analyzed. The key process to calculate the area of polluted water is to identify the temperature threshold that distinguishes polluted water from clean water. To solve the difficult, the artificial interpret method and the differential approach were proposed. After analysis, it was found that although there was some difference in the calculated areas of waste water, the two algorithms could be used to estimate the areas of the sewage objectively. The artificial interpret method was flexible but much more subjective, while the differential approach was more objective but weak in determination of the edge of different waters. When in applications, the algorithm should be chosen according to the actual situation.