Evaluation of definition for gray scale digital image is an important aspect of digital imaging system. Thus

in order to evaluate the definition of a gray scale image accurately and effectively

we present a new approach based on EVA method which

while retaining important features of existing method

overcomes some of their limitation. With curvilinear regression analysis we can prove that the result of the new approach is high correlation with MTF measured by optical instrument

in other words

the new method is sensitive to the change of image definition. When compared with traditional method

the result of analysis can also show that the new method is better than traditional method

such as entropy

to assess gray scale digital image definition. Experiments using hundreds of many kinds of gray scale digital image and the result of this new approach is well accurate to the change of definition of digital image. Form these we can also draw a conclusion that the new approach can be well applied to many kinds of gray scale digital image accurately and effectively.