Real time traffic occurrence recognition is decisive for mounting safety and mobility on freeways. Numerous incident recognition approaches are presented based on traffic activities or geometric models proposed for better sequential mining tasks. Nevertheless, former incident recognition methods are restricted in distinguishing persistent and non-persistent congestions. The difficulty of present techniques makes them inadequate to hold the genuine time task. The previous work described the process of identify similarity among two sequences of finger print images alter in time or speed can be applied to video, audio, and graphics with linear representation. In the second phase of the work, evaluation made on spatial and temporal information are measured to identify the possible incidents. In this paper, a novel approach for identifying incidents is proposed. Unlike from conventional traffic occurrence recognition methods, both spatial and temporal information are measured to discover the probable incidents. In the meantime, adaptive learning capability and short recognition response time are attained in the novel method. To examine the high dimensional transfer data, Mahalanobis distance is processed to determine possible incidents consistent with the traffic outline. Lifeline style recognition and revelation is employed to present instinctive user interface. Methodology examination and preface assessment have been achieved to authenticate the recognition efficiency on the integrated traffic revelation system. A real-time spatial-temporal pattern mining technique, especially for the data traffic and fast response time is attained by utilizing active identification approach. A set of tests have been conducted in real system to validate the effectiveness and efficiency of traffic occurrence detection using spatio temporal data mining approach [TODSTD].
