Statistics for Recognition of the Script in Serbian Documents using Frequency Occurrence and Co-occurrence Analysis