Your question seems to indicates you do not understand how IP works, how IP routing works and how TCP is implemented on top of it (and incidentally, what IP address spoofing really means)
IP implements a routed datagram protocol: each datagram (packet) has (among other things) a source address, a destination address and a payload (there are many other properties but they do not impact this discussion).
In the simplest case (i.e. without any type of filtering or NATing in place), each router along the way only looks at the destination address, match it against its own routing table and then forward it to the next hop on the correct interface. The return address is only looked at by the system at the end of the path (either the destination host or, if the packet has been dropped early, the system that needs to send the notification back).
Spoofing the source IP means replacing the source address of a packet by some other random host. It is usually (not exclusively) used in order to hide the source of this packet, to force the target into sending network traffic in direction of the spoofed host (typical of a network traffic amplification attack like DNS amplification).
Detecting a spoofed packet can, therefore, only be done close to the source of the traffic: In a simple case, the first router on the path has the possibility to detect that the source address in the packet does not belong to any of the internal networks it knows of and therefore could drop it (that is called "egress filtering"). Unfortunately, this type of check can only be done inside or at the edge of a network and it is usually only performed by firewalls and needs to be carefully setup to avoid side effects (and therefore, not widely implemented).
Another almost identical technique, ingress filtering, tries to packet coming into the network but it must rely on some knowledge of the connected network. Typically, it only works of the filtering devices knows of all the networks that are connected through it extensively (i.e. it's mostly useful between members of a peering).
TCP implements a stream connection on top of the IP protocol. As part of this implementation, it will require multiple IP packets to be sent back and forth between the client and the server in order to exchange any data (first through a 3-way handshake to establish a "connection" and then through packet acknowledgment). It means that, should the source address of an IP packet of type TCP be incorrect, the peer will not be able to send the proper response to it and the packet will be dropped.
In some special cases, it is still possible to spoof a source address in a TCP connection but that requires the source of the traffic to either be able to monitor or predict the response from the target host in order to forge the proper answer.
In regard of the above:
How is it possible to find out the address of spoofed IPs?
Assuming you mean "the real address of the spoofed packet", you can't unless you're on the same subnet as the source. Any packet that is forwarded any further will lose more and more information about the real source of the traffic (you will only be able to know accurately which interface the traffic arrived from on and therefore potentially limit the potential networks it could have originated from, but nothing else).
What methods are used?
The only way to find the source of a spoofed packet is to monitor each hop on the path of that packet and identify where that traffic comes from. Then, you do the same with this source until you've narrowed the potential source enough. This typically requires the cooperation of each and every hop on the real path of the traffic and that is a slow (and expensive) process. It's typically not worth the effort.
How do you detect spoofed IPs?
It depends on what part of the network path you're talking about.
On the final host, it is very difficult: you more or less have to have a way to vet for each and every packet that arrives (for instance, by preventing datagram traffic coming from external networks to reach your host and only allowing TCP or similar protocols.
At or near the source of the traffic, it is much easier: you simply drop every packet that has a source address that isn't part of the networks you know are connected to a specific interface.
In between, you could filter traffic using in/egress rules (although the further from the source of the traffic you get, the less efficient that becomes).