The article presents new results in studying the problem of detecting sharp changes in generally smooth (piecewise constant type) signals. These results were obtained based on the previously developed neuromorphic approach to processing streaming data. The article also provides a brief overview of the proposed approach. In particular, a description of the sampling representation (model) of a piecewise constant intensity Poisson point signal (process) is given. Based on this model, a statistical procedure for encoding intensity changes is developed. Theoretical analysis of the procedure is central to the work. The results of the theoretical analysis of the procedure are implemented in the article in the algorithm for detecting change points in the intensity of Poisson point processes. Some features of the obtained algorithm are discussed in the article and presented in the conclusion.