Thermal camera has been applied in photovoltaic (PV) array monitoring to detect defects and hot spots. However, the infrared (IR) image is often degenerated and the temperature information displayed from the image can lead to erroneous interpretation. Hot spots will be obscured and a gradual change phenomenon will emerge when monitoring a large PV array. In this paper, the mechanism of IR image degeneration and gradual change phenomenon are studied and verified with experiments. The variations of atmospheric transmission and directional emissivity have been identified to be the cause of image degeneration. The sensitivity of atmospheric transmission and directional emissivity are defined to analyze the impact of these two factors on the temperature displayed. Based on this mechanism, a recovery method has been proposed to recover the real temperature from the degenerated IR image. Experiments have been conducted to test the effectiveness of the recovery method. In addition, the temperature sensitivity has been defined to analyze how atmospheric transmission and directional emissivity will affect the temperature difference displayed in the IR image. The proposed temperature sensitivity has been used as the criteria to assess the quality of IR image. Some rules of thumb are proposed to deploy the thermal camera in order to increase the quality of the IR image.