Texture is one of the most important characteristics for the substance recognition both in visual perception and computer image analysis. Many texture models have been used. But whether the model is appropriate in mathematics has not been studied before. In this paper

a test of hypothesis which examines the autoregressive texture model in introduced. The test of hypothesis is also used in looking for a more appropriate autoregressive texture model.In this paper

the autocorrelation method is also suggested to reduce the computational workload.