In order to analyze theory and method of coal level detection by intelligent camera, firstly, mathematic model of coal level detection in bunker was established by three-dimensional imaging and synthesis and decomposition of vision signal, which is based on Faugerast�s calibration method, and the variation law between image signal and coal level variation in bunker were obtained by theory analysis. Secondly, the experimental simulation system for coal level detection in variation induced by the variation of image signal was constructed, and the image signal variation law was revealed. The numerical simulation model of the coal level detection in vision display was established, the simulation show that the greater the peak and contour points of coal level are, the faster the variation of image signal is, and the greater the initial value of coal level is, the higher the image signal is, and the coal level may have lightly pit. Finally, the image reconstruction shows that the theoretical analysis, experimental research and numerical results are consistent.