The use of higher-order local autocorrelations as features forpattern recognition has been acknowledge since many years, but their applicability was restricted to relatively low orders (2 or 3) and small local neighborhoods, due to combinatorial increase in computational costs. In this paper a new method for using these features is presented, which allows the use of autocorrelations of any order and of larger neighborhoods. The method is closely related to the classifier used, a Support Vector Machine (