considering the facts that SHA256 transforms 256bit public key (as input) to 256bit output, but RIPEMD160 transforms 256bit to 160bit, so the highly likely the collision is in RIPEMD160 function.
That depends on what you mean by
highly likely.
It is much MORE likely that it will occur in the RIPEMD160 transform than in the SHA256 transform, but that's a bit like saying that it is much MORE likely that a completely fair coin will land on heads 160 times in a row than 256 times in a row. In both cases it is *VERY* unlikely.
Essentially we are talking about the difference between "It isn't going to happen" and "It isn't going to happen".
I used the incorrect words to descripbe. Of course the probability is highly unlikely, it is actually impossible. And for me as 2^256, so 2^160 are both very high numbers, and the probabilities 1/2^256 and 1/2^160 are actually 0%.
I just had in mind the probable place of collision if such collision is found. Let's say Key1 and Key2 are 2 keys leading to the same Address. So, it is more likely that public keys PubKey1 and PubKey2 are different. Then it is also more likely that SHA256 of PubKey1 and SHA256 of PubKey2 will be different. But the collision is in RIPEMD160 function transforming different SHA256(PubKey1) and SHA256(PubKey2) to the same hash, and then to the sme Address.