區塊式濾波反投影演算法設計與FPGA實作(I)

Abstract

濾波反投影(filtered backprojection)演算法在電腦斷層掃描扮演相當重要的角色，其計算繁瑣及複雜。我們目標設計一個可高度平行化處理的演算法，由於硬體資源不足，無法對整張filtered sinogram做影像重建。因此採用區塊式重建，逐個區塊做影像重建，即為最後結果。經由軟體模擬，模擬結果有大小不同的區塊，我們都得到預期的結果。此論文需參照林坤世同學所撰寫的＂區塊式濾波反投影演算法設計與FPGA實作(II)＂。

The filtered backprojection algorithm plays an important role in computational tomography, its calculations are complicated and tedious. All we want to do is to design a highly parallel algorithm to solve the limitations of hardware resources. Because of lack of resources, we take the block-based policy, doing reconstruction block by block. After the final block has been reconstructed, the block-based FBP is done. By implementing the algorithm we design, we have the expected consequences in different block size. You should refer to the thesis “Block-based Filtered Backprojection Algorithm Design and Implementation on FPGA (II) ” written by Kun-Shih Lin.