Ubiquitous monitoring has become a useful tool for the prevention and early diagnosis of some disorders. This kind of monitoring is nowadays very common thanks to the irruption of smartphones, which make easier the collection and delivery of the patient’s data. However, some problems arise due to the occasional low quality of the acquired data. This paper presents a novel methodology in which the shape of PPG beats is recovered through a multistage fully-automatic pipeline featuring a shape modeling stage as well as a template-based shape recovery through a level-set approach; the template is not assumed or derived from previously acquired data but it is estimated from the surrounding beats. To validate the proposal, a registry acquired using the camera of a Motorola MotoG has been used. Results can be qualified as promising, since the shape of damaged beats is recovered, although a comprehensive validation should be addressed.