Image Registration Based On Phase Correlation Algorithm And Its Application In Ultrasonic Testing

New study shows an increase of cardiovascular disease and cancer risk for younger groups of people:incidence, prevalence, and morality. Those have become most serious diseases threatening human health. Ultrasound examination technology can retrieve physiologic and anatomic information, to help disease detection, prevention and control. As a non-invasive technology, it has the distinctive importance in disease diagnoses for such as craniocerebral disease, pleural disease, digestive system disease, genitourinary disease.Real system environment may not be ideal:device vibration caused by operational mistake, patient breath will introduce a large margin of error. Image processing has been used to correct the variation, while currently most study is about Matlab registration algorithm optimization and there is no image processing system available specifically for error-correcting. To process hundreds of ultrasound images is time-consuming. Phase correlation has been the common and widely used image registration technology with relatively strong noise cancellation performance and high robustness. It relies on image gray scale and is less influenced by ultrasound image speckle noise. By converting images to frequency domain, image features become more obvious for extraction. However most ultrasound image registration researches are focusing on non-linear registration schemes for soft tissue, and there hasn’t been an approach to apply rigid registration "phase correlation" in displacement detection for vascular wall ultrasound images.With the aforementioned two issues, this document designs a phase correlation image registration system for ultrasound images noise detection. The system supports multiple registration algorithms, including translation, rotation and scaling, supports multi image batch processing. Has dramatically improved processing efficiency and operational flexibility. Finally compares the blood vessel vibration result images from MATLAB simulation, with output analysis proves that the system has relatively accurate registration results, and meets original performance expectations.