In this paper the results of different feature extraction algorithms are used to build feature-lists. These feature-lists are used for motion estimation in retinal fundus image series. Therefore the feature-list cross-correlation algorithm is used. The influence of the feature extraction on the results of the feature-list cross-correlation is evaluated. Therefore different kind of image series, with different kind of selected templates is used. The reference position is determined by the median of the detected position of all templates and all feature extraction algorithms. The amount of incorrect detected templates is compared. The Harris corner detector detects only the optic nerve sufficiently. The Sobel-operator delivers the best results, except using small templates. The Canny-edge-detection has good results too. Over all, the rule-based edge detection delivers the best result. Generally, it is possible to use different feature extraction algorithms to estimate motions in retinal image series by using feature-list cross-correlation algorithms. As long as the amount of feature values of the template or the image is much smaller than the amount of pixels in the image, the feature-list crosscorrelation algorithms are faster than the common cross-correlation algorithms.