To encode the signature you first have to parse the ASN.1 structure also shown in the linked answers provided by dave_thomson_085. Once you have got two integer in the programming language / runtime of your choice then you need to encode them as two statically sized integer encodings. These integers (usually) need to be big endian / network order, unsigned integers.
To do this you need to implement a function called I2OSP from the RSA specifications. You feed this the two numbers and the key size (the size of the order of the curve) rounded upwards to bytes (that's $192 / 8 = 24$ bytes).
There are two ways to do this:
- directly implement the mathematical functions and concatenate the resulting bytes or
- encode the integer back to bytes - most platforms have a function for this - and then adjust the encoding by padding and possibly reversing the encoding
The latter is probably faster as it just requires byte operations. I usually opt for the latter, but I've seen many cryptographers (naively, in my opinion) do the first.
Finally you simply concatenate the numbers and presto - you're done (there is also OS2IP in RSA, in case you need to do the reverse).