The estimation and the controlling processes in the crowd counting filed are the most important on over applications related to surveillance systems and control follow related peoples to give them a complete safety especially in the large groups that contain a huge collection of individuals. There were found various events in human histories that gather a huge number of crowds in the same place. For example, in religion occasions as in HAJJ season which occurs every year, these events cause accidents that lead to death. So, to avoid and prevent all types of accidents related to these huge crowds; this paper presents an improved technique to estimate the density of peoples in the same place to help the decision makers to monitor and control the pedestrians overcrowded in the large collection of individuals. The proposed technique depends on the thermal bands for human, according to the big variance of temperature human body between the skin and the skin covered with clothes. Also, it presents the whole range of the temperature for each frame in the video. The essential characteristic of crowd counting technique is that it does not require a previously stored and trained data, but it uses a live video stream as input. Also, it does not require any intervention from individuals. The research's approach depends on capturing the thermal features of an individual. The result of this technique is introduced and proved to be highly accurate, and the experimental results demonstrate the effectiveness of the approach.