<p>Rock mass discontinuities play a crucial role in controlling the stability of a slope. They are frequently in the terrain of a high steep slope, which makes discontinuities challenging to be observed and recognized. Visible images (RGB images) rely heavily on illumination that leads to uncertainty shadow or backlit effect, and the inherent color difference of the rock masses also pose a challenge for the computers when recognizing the exposed linear discontinuities. To address these concerns, we conduct the research on multi-source investigation for the exposed linear discontinuity recognition on a high steep slope using the unmanned aerial vehicle (UAV). First, both thermal infrared and visible imaging technologies are utilized to monitor the exposed rock mass surfaces and the appropriate time is explored to capture the high-quality visible and thermal infrared images simultaneously. Second, a novel method is proposed to fuse the multi-source images, which can eliminate the noise from the perspective of the single channel and improve the fusion efficiency. Through the multi-source image fusion, the exposed linear discontinuities are significantly visible and more discontinuities on the exposed rock mass surfaces can be recognized using the proposed computer vision algorithm.</p>