Recognizing and processing documents, especially historical manuscripts, has been a challenge owing to the difficulty in identifying cursive handwritten characters. This paper presents a simple technique for Optical Character Recognition based on Swarm intelligence to extract textual information from printed or handwritten documents. Primarily, combinations of different filters to remove noise and background stutter is used. After employing frequency domain transforms for feature extraction on the image containing the textual data, the Discrete Artificial Bee Colony algorithm (DABC) is used for selecting useful features. Experimental results show that DABC based feature selection method provides good recognition by removing noise and redundant features. The proposed algorithm works well not only with machine-generated characters, but also with handwritten characters of multiple languages including English, Hindi and Kannada. The technique is comparatively more complete as it works for both machine generated and Handwritten databases, and also aims at achieving language independence.