Image-quality criteria are usually intended to achieve the best possible image at a given sampling rate, which is ill-suited to applications where the detection of well-defined geometric and radiometric properties of scenes or objects are paramount. The quality criterion developed here for designing observation systems is based on properties of the objects to be viewed. It is thus an object-oriented imaging quality criterion rather than an image-oriented one. We also propose to go beyond optimization and calibrate a numerical scale that can be used to rate the quality of the service delivered by any observation system.