Grade judgment is an essential component of a vegetable and fruit sorting system. Judgment of capability should be integrated with the collection of raw data from an object, and classification of algorithm of processed data. Machine vision is a useful tool for collecting raw data and for classification. Recently, machine vision applications for sorting and inspecting some fruit vegetables have been studied by many scientists ^{4-7, 10, 12-16}), However, the traditional image recognition and image understanding methods cannot evaluate satisfactorily irregularly shaped fruit vegetables. Thus, in this article a new feature extraction procedure and neural network were applied to the machine vision system to precisely evaluate fruit vegetables. As the newly developed machine vision system has a learning capability, it can evaluate more than one variety of fruit vegetables. The results obtained for 4 kinds of strawberry varieties and 1 variety of green pepper showed a high accuracy, confirming that the system has a great potential for grade judgment ^{3, 9}).