Abstract:  On the basis of improving the performance of traditional vehicle targets detection method，this paper focuses on effectiveness and accuracy，and proposes a new method of vehicle targets detection based on morphology top­hat transformation combined with pulse coupled neural network(PCNN)．First，morphology top­hat transformation is utilized for traffic image to extract the morphology edges of all objects.Then pulse coupled neural network(PCNN)is utilized to analyze the discrimination between vehicle morphology edge image and non­vehicle morphology edge image.In the process of analysis，we evaluate the effectiveness of entropy features(EF)and pulse firing characteristics(PFC)on both the original image and top­hat image，respectively.Iterative average entropy features(IAEF)is chosen as the best characteristics for vehicle targets detection according to a large number of image statistics，and top­hat image is more suitable for vehicle targets detection.We take advantage of iterative average entropy features(IAEF)to detect the vehicle targets through a sliding window，and the window size is adaptively determined according to the test images.Finally，the edge density information(EDI)is utilized to confirm the detected vehicle targets，only high edge density information images can be recognized as a vehicle image.The experiments in this paper are verified from two aspects of validity and accuracy，and all test images are come from different actual traffic junctions in Wuhan.The experiments show that the method proposed in this paper can effectively for vehicle targets detection.Compared with other vehicle target detection methods，our method has higher detection rate and lower false detection rate and less time，and can better realize the goal of rapid vehicle targets detection in the intelligent transportation system(ITS)with brilliant prospects in applications.