Synthetic aperture radar (SAR) interferometry is a powerful and consolidated technology to detect and monitor slow ground surface movements. However, information extraction and interpretation from big sets of InSAR deformation measurement is still a complex and demanding task. In this paper, we present a new approach for automatic analysis and detection deformations anomalous in the spatial or the temporal dimension. The method takes in input InSAR deformation measurements. In the spatial dimension, in order to reduce the dataset size, we utilize a hierarchical clustering method to obtain convergence points which are more trustful, and then to detect spatially inhomogeneous deformations. In the temporal dimension, we use a signal processing method to decompose the input into two main components: regular periodic deformations and piecewise linear deformations. After removing the periodic component, the PS velocity variation in each identified homogeneous period is analyzed to detect anomalous velocity trends and accelerations. We have tested the method on different cases, and we report the results obtained on PS measurements from high-resolution X-band COSMO-SkyMed data on some sites in China. The proposed method appears very promising, as confirmed from the in-field survey, in enabling the detection of deformation anomalies that could cause building or infrastructure stability problems, and in some cases could bring to collapses and potential disasters.