Abstract: The threshold method is commonly used to determine cloud in sky, which has great limit when it comes to thin cirrus. This paper is focused on improving the ability of distinguishing high or thin cloud from sky. Based on the developed cloud detection by all-sky infrared and visible imagers, a new modified method is presented. Measurements of sky visible image and thermal infrared brightness temperature were conducted at the top of the laboratory building, using an automatic visible all-sky imager (ASI) and an automatic scanning infrared thermometer (SIRIS) developed by the Key Laboratory of Middle Atmosphere and Ulobal Environment Obscrvation (LAGEO),the Institute of Atmospheric Physics (IAP),Chinese Academy of Sciences. hhe characteristics of pixels which arc symmetrically disposed by the solar principal plane arc used to estimate the existence of cloud in a visible image. As for SIRIS, the heteropic distribution of brightness temperature (TB) from the same elevation and the different azimuth can tell whether there is cloud. The cases at Beijing during the period January to March 2011,in which clouds do exist yet cannot be detected before are sorted to test the algorithm. The result shows that the ability of distinguishing high cirrus for ASI has reached to 73.9% and for SIRIS has reached to 70.1%.This work provides a new composite algorithm for the high cloud recognition.