Locating Straight Image Features to Sub-pixel Accuracy

T. Bretschneider, O. Kao, and G. Joubert (Germany)


point-spread function, locating features, sub-pixel accuracy


This paper presents a technique to locate straight image features to sub-pixel accuracy using under-sampled imagery, whereby it is assumed that the approximate lo cation was previously determined by an arbitrary edge detection filter. The approach requires the a-priori knowledge of the acquisition system’s point spread func tion and computes the solution in a closed-form by com paring the actual feature with its blurred model. The in corporation of the system characteristic proved to result in more accurate estimates for the feature locations than other techniques. Moreover it is insensitive to noise. One of the major advantages of the proposed technique is that it also enables the use of not only features which are best modelled by a simple step edge but also more complex edge models. Possible applications for the proposed method are to be found in mapping, robot vision etc.

Important Links:

Go Back