This paper proposes an image encryption method using a spatio-temporal cryptographic diversification strategy. The implemented method, called Pseudo-Random Image Encryption System (PRIES), is based on three conceptual components: What (choice of methods), Where (spatial segmentation) and How (temporal assignment). The What component uses a cryptographic repository of image encryption methods (CRIEM) to pseudo-randomly provide, at each encryption cycle, a sequence of several different methods. The Where component employs a radial segmentation strategy inspired by the wavelet domain, segmenting the image into various angular segments to break spatial correlations. Finally, the How component implements a Markov model to pseudo-randomly assign the selected sequences, defined in the What component, to the angular segments defined in the Where component. On the operational level, experimental data demonstrate high robustness and cryptographic performance: an entropy close to 7.999, a pixel change rate (NPCR) greater than 99.6%, a uniqueness of pixel change index (UACI) approaching 33.46% and a very reduced processing time.