Improving accuracy and speeding up Document Image Classification through parallel systems