In detecting QRS peak, the most easiest thing is detecting the R peak , if you use wavelets. Once the R-R peaks are found then Q & S won't be much problem , since Q ,R and S are related among each other with certain proportions. Wavelets is a good tool for this purpose. I have worked on it earlier some time back.
I strongly feel that ECG does'nt have much for human detection because in every human heart beat varies in accordance to situations, mood and other factors, so the distance of peaks varies, which then breaks the property of an independent feature. Maybe u can try some other parameters along with it still , it won't help much.