A method based on statistical modeling of the spatial structures of the images to identify the product-quality is presented.In advance,the image spatial structures are proved to be subject to a typical Weibull distribution (WD) theoretically.A method of multi-scalar and omni-directional Gaussian derivati ve filtering is adopted to extract the spatial structure details of the monitoring images.Hence,a signific ant omni-directional statistics feature vector of the filtering response based on WD model with various observ ation scales is obtained.At last,a least square support vector machine (LS-SVM) based product-quality classifier is established to identify the production quality automatically.The proposed method is tested in the fields of automatic i dentification of the rice production-quality and intelligent assessment of the surface quality of the gre ige cloth in the corresponding assembly lines of industrial scale.Experimental results indicate that the propo sed image statistical modeling method can effectively characterize the statistical distribution of the spatial structure of these complex texture images piled with a large number of local debris particles randomly.Hen ce,the automatic classification model can achieve high accurate rate.The proposed method lays a foundation for the intelligent perception of the product-quality and the automatic processing of the high qual ity products in the assembly lines.