文摘
A new machine vision approach for estimating, monitoring, and controlling manufactured productappearance is illustrated. This new approach consists of the following: (1) extraction of texturalinformation from product images, (2) estimation of measures of the visual quality of the productfrom the textural information extracted, (3) modeling causal relationships between the estimatedquality and process variables, and (4) optimization of new operating conditions using the causalmodel. This method is specifically aimed at treating the stochastic nature in the visual appearanceof many manufactured products. This nondeterministic nature of product appearance has beena main obstacle for the success of machine vision in the process industries. This approach issuccessfully applied to an industrial process for estimation, modeling and optimization of thevisual appearance of injection-molded plastic panels.