The license plate detection is one of the most important processes in license plate recognition (LPR) system. A number of techniques have been proposed in research literatures. Among the most promising techniques is text locating-based technique. However, in a complex scene the technique creates too many candidate regions with noisy background which result in segmentation difficulty. Toward the text locating approach, this paper presents a new license plate detection and segmentation technique. Improving on the existing technique, it creates cleaner connected regions by using our new morphological approach. The searching and segmentation technique proposed here is also efficient for this very application.