Abstract: For better simulating the dynamics of leaf area index (LAI), an experiment under field conditions, with five varieties and five densities at different stages, was carried out. The digital images of wheat groups canopy were obtained, , and the actual leaf area was manually measured. An effective extraction method of wheat LAI under complex background was designed. Simulation models between data of image processing and actual leaf area were established. Results showed that the model parameters were significantly affected by the differences of variety, density and growing stages. Hypothetical tests of the models were carried out by random extracting sample images, and all of them were able to meet the requirements of test. Square of the correlation coefficient of all models is higher than 0.86. This can achieve the higher precision estimation of wheat canopy LAI.