Compared with its counterparts, polarization sensitive optical time-domain reflectometry (P-OTDR) has two typical problems in terms of multi-point disturbance detection and high precision positioning, which restrict its practical applications. In this paper, we propose a novel method by using two-dimensional (2-D) image processing and statistical clustering. In place of the traditional spatial or temporal signal analysis way, the idea employs the temporally and the spatially uneven/inconsistent evolving patterns induced by the multiple individual events, and the proposed method is carried out to accumulate the temporally differentiated OTDR traces and then build up a 2-D temporal-spatial evolving graph, by using edge detection and automatic clustering to distinguish different disturbing points with precise locations. Through series of multi-point disturbance tests, the proposed method has been proved to have better performance than the conventional direct differentiation method and the fast Fourier transform spectrum analysis.