摘要
在临床上,使用多种成像技术并适当地将其加以融合,可为临床诊断和手术治疗提供更加全面准确的信息。而图像配准是融合技术中需要先期解决的问题,也是融合的关键部分。本文提出了一种基于DSP的医学图像配准问题的研究与实现的方案。该方案首先采用TI公司的高性能多媒体数字信号处理器DM642获取实时医学图像,再用配准算法验证其获取的实时医学图像配准的可行性。并采取了基于轮廓提取和最大互信息的方法进行图像配准,克服了互信息单一的利用图像灰度信息的局限性,将边缘检测与互信息相结合,较准确地完成图像配准任务。
Clinically, the use of a variety of imaging techniques and where appropriate, to integrate its for clinical diagnosis and surgical treatment to provide a more comprehensive and accurate information. Image registration and fusion technology is the need for early settlement of the issue, but also a key part of integration. In this paper proposes a DSP-based medical image registration research and implementation of the program. The program began using Trs high-performance multimedia DM642 digital signal processor to obtain real-time medical image registration algorithm and then verify its real-time access to medical image registration is feasible, we have taken and the largest based on the contour extraction method of mutual information image registration, mutual information to overcome the use of a single image of the limitations of the information, will edge detection with a combination of mutual information, more accurate completion of image registration task.
出处
《齐齐哈尔大学学报（自然科学版）》
2011年第1期35-38,共4页
Journal of Qiqihar University(Natural Science Edition)
关键词
图像配准
最大互信息
边缘提取
image registration
maximum mutual information
edge detection
