Change Point Detection in time series aims to identify the time points when the statistical properties of a time series change. A major challenge in this field is how to reduce the effect of noise in the time series on the accuracy of change point detection. This is particularly challenging when the time series exhibits complex patterns of normal behaviour. Recently, Topology Data Analysis has shown the potential to characterise the shape of complex and irregular time series. In this paper, we propose a method called Smoothing-based Change Point Detector (S-CPD) (Source code is available here: https://github.com/harinduashan/S-CPD .) which transforms the time signal to a smoothed signal by using the lower-star filtration-based Persistent Homology from the field of topological data analysis. We then detect the change points in the transformed signal by using a classification-based score profiling. The benefit of S-CPD is its ability to interpret noisy information as low-persistence topological data and remove it, allowing the refined signal to be effectively used for detecting change points. We demonstrate the effectiveness of our method on two real-world datasets in comparison with several state-of-the-art change point detection methods.