What exactly is the format used for encoding? ASCII encoded in base58 format (minus the version, and the checksum)?
Bitcoin address in its raw format contains 20 bytes of data. These 20 bytes are later converted into base58 and a checksum is appended, thus we get a typical bitcoin receiving address. To decode an address I restore the underlying 20 bytes from the bitcoin address and seek common ASCII characters from it. If more than 90% of the characters are visible ASCII characters then the address is assumed to be human constructed and thus is probably a human readable text.