Merging mate pairs by quality

[Yrinky]

Hello
First I want to apologize if there's a similar question already, but I can't seem to find an answer here.
I have to fastq files, one containing forward, one reverse mate pair and they are in the same order in both files. Each of those pairs overlap and what I want to do is to somehow merge them into one sequence to get a consensus sequence based on their qualities. These are Sanger sequences and are a bit longer.
Is there any software or a script that I could use for that? So far I've tried CAP3 and didn't manage to get desired results.
Thanks in advance.

[maubp]

So this was a small fragment paired end sequencing run, say fragments of 150bp and reads of 100bp, which overlap by about 50bp - and you want to turn each 100bp forward + 100bp reverse read into a single 150bp read? I've seem something for this but the name escapes me...

[Yrinky]

The fragments are longer, about 500bp and from what I've seen for manual pairwise alignments they are mostly overlapped. Reverse reads are already reverse complemented so they just need to be merged.