Panoramic imaging of ultrasound videos based on multi-mixing measures

This paper presents a new panoramic imaging algorithm. With the target being direct video data, it could be absolutely applied in the actual ultrasonic application. In order to reduce the computation, this paper chooses sampling frames from the video for panoramic imaging. Therefore, the algorithm is actually a cyclical sampling process. The loop includes five parts: sampling from video data, evaluation of sampling frames, registration, adjusting sampling steps and splicing. During the whole process, different kinds of measures are applied for panoramic imaging, including SAD (Sum of Absolute Difference), statistical measure, MI (Mutual Information) and GC (Gradient Correlation). The application of multi-mixing measures provides a strong robustness and a high precision for the algorithm. Besides, the sampling process greatly reduces the computation. With adaptive sampling steps, the algorithm is able to meet the real-time requirement of ultrasound systems. By simulation and statistical data, the algorithm has been proven practical in clinic.