基于图像分割的Tamura纹理特征算法的研究与实现

Research and realization of Tamura texture feature extraction method based on image segmentation

背景:ITK主要提供医学图像处理、分割与配准算法,但其缺少可视化的功能,缺乏灵活实用的用户界面,VTK提供了丰富的医学影像处理与分析工具,具有强大的图形处理和可视化功能。目的:利用以前的确诊病例和医生的诊断经验以及患者的相关病史,对确诊的医学影像资源进行管理,归档,并检索,以减少人工干预,提高图像的查全率和查准率。方法:以视觉感知机制为基础,在ITK平台上进行图像平滑去噪和分割的预处理过程,利用Tamura算法完成纹理特征提取,最后通过实验采集、计算数据,完成对比分析。结果与结论:基于图像分割的Tamura纹理特征算法在基于图像纹理检索应用上便于相似性度量,进而可提高检索的准确率。

BACKGROUND：Insight segmentation and registration toolkit is the main provider of medical image processing,segmentation and registration algorithm,but,it lacks of visual function and flexible and practical user interface.Visualization toolkit provides a wealth of medical image processing and analysis tools,with the powerful graphics and visualization capabilities.OBJECTIVE：To manage,archive and retrieve the diagnosed medical imaging resource by using the previously confirmed cases,diagnose experience and relevant medical history in order to reduce manual intervention and improve the image recall rate and precision rate.METHODS：The images were preformed with smoothing,de-noising and segmentation pretreatment process on the insight segmentation and registration toolkit platform based on visual perception mechanism,and then texture feature was extracted by Tamura algorithm.Finally,the comparative analysis was completed through experiment data collection and calculation.RESULTS AND CONCLUSION：The experiment has proved that the method based on the application of texture feature retrieval can facilitate the similarity measurement and improve the retrieval accuracy.