Comparison of Interpolation Methods in Medical Image Processing
Haiping Ren · Beijing shengwu yixue gongcheng · 2002
The technique of image interpolation is usually required for medical image processing, analysis and display. Programs were performed on PC to simulate the three main situations (resolution change, image rotation, view angle change) typically required in medical applications. In this paper, results obtained by several scene-based interpolation methods, such as truncated and windowed sinc, nearest neighbor, linear, cubic, B-spline and Lagrange polynomial interpolation were compared. In each case, the efficiency and accuracy of a particular interpolation technique were evaluated by analyzing its visual quality, statistical interpolation error and relative runtime. Summarizing the evaluations, cubic and sinc-cosine interpolator provided better results. When using small interpolation operator, it may be considered to use the 2×2 cubic interpolator instead of the linear interpolator most often used currently.