點雲為三維空間中無序的點集合，透過掃描可以獲取包含顏色、位置等數據。點雲在醫學建模、自動駕駛導航等領域廣泛運用。然而在獲取過程中，因傳感器誤差、環境干擾、或操作等因素可能對點雲數據產生幾何和顏色上的影響，降低其原始質量。因此，本研究提出了一種應用於積體電路設計中受到高斯雜訊影響的三維點雲平滑的去雜訊演算法。該演算法分成兩個階段，首先對目標點做k-nearest neighbor (KNN) search找到離目標點最近的鄰近點，根據這些鄰近點的三維座標替換目標點原先的座標，從而實現對受雜訊干擾的目標點進行低複雜度平滑處理。此階段探討均值濾波器和中值濾波器的功效，並通過對比和分析結果，最終選擇均值濾波器作為第一階段非線性濾波器。該選擇不僅對後續的硬體實現更為理想並提供更好的效果。第二階段是一個改良版的雙邊濾波器。它使用兩個權重來進行過濾，首先計算點與點之間的距離作為第一個權重，然後計算點的法向量夾角作為第二個權重。這樣的操作能夠更好地抑制雜訊並有效地保留邊緣資訊。根據本研究提出的演算法讓峰值訊噪比（PSNR）相比僅使用雙邊濾波器提升了24%，均方誤差（MSE）的平均值降低了90%。
本研究提出了一種針對雙邊濾波器權重計算的簡化方法，以降低其複雜度並保持輸出品質。在雙邊濾波器的計算過程中，需要進行大量的乘法運算，包括點之間的運算和法向量相關運算。為了解決這些複雜運算的問題，本研究設計了一個稱為boost encoder的乘法器，用於執行這些運算。該乘法器利用移位和加減法運算的方式，有效地減少了對乘法器的需求。對於未來的研究，可以著重於進一步優化電路設計，以提升計算效率和性能。同時，也可以將電路設計的應用領域擴展到更廣泛的領域中，以達到更廣泛的應用範圍。

Point clouds are an unordered collection of points in 3D space that captures data including color and position through scanning. Point clouds have found widespread applications in various fields, such as medical modeling and autonomous vehicle navigation. However, during the acquisition process, factors such as sensor errors, environmental interference, and improper operations can introduce geometric and color noise, leading to a degradation of the original quality of point cloud data. Therefore, this study proposes a denoising algorithm for smoothing 3D point clouds affected by Gaussian noise, specifically designed for integrated circuit design. The algorithm consists of two stages: first stage employs a non-linear filter that utilizes a k-nearest neighbor (KNN) search to find the nearest neighboring points to a target point and replaces the coordinates of the target point with those of the neighboring points, achieving low-complexity smoothing for noise-affected target points. In this stage, the effectiveness of median and mean filters is investigated, and based on the comparison and analysis results, the mean filter is chosen as the first-stage non-linear filter in this study, which not only facilitates subsequent hardware implementation but also provides superior performance. The second stage introduces an improved bilateral filter that calculates the distance between points as the first weight and the angle between the point's normal vectors as the second weight. This operation effectively suppresses noise while preserving edge information. The proposed algorithm demonstrates an improvement in peak signal-to-noise ratio (PSNR) by 24% compared to using only the bilateral filter, with an average reduction in mean squared error (MSE) by 90%.
The VLSI circuit design proposed in this study simplifies the weight calculation of bilateral filters, reducing its complexity while maintaining excellent output quality. The computation of bilateral filters involves a significant number of multiplications, including operations between points and computations related to normal vectors. To address this issue, a boost encoder multiplier is designed in this research to perform these computations. The design of the multiplier utilizes shift and add/subtract operations, effectively reducing the demand for multipliers and improving efficiency. This contribution holds significant value for the practical application of bilateral filters in VLSI circuit design. Future research can further optimize circuit design to enhance computational efficiency and performance, extending its application to a broader range of domains.

[1] R. B. Rusu and S. Cousins, “3D is here: Point Cloud Library (PCL),” in 2011 IEEE International Conference on Robotics and Automation, Shanghai, China: IEEE, May 2011, pp. 1–4. doi: 10.1109/ICRA.2011.5980567.
[2] C. Tomasi and R. Manduchi, “Bilateral filtering for gray and color images,” in Sixth International Conference on Computer Vision (IEEE Cat. No.98CH36271), Bombay, India: Narosa Publishing House, 1998, pp. 839–846. doi: 10.1109/ICCV.1998.710815.
[3] S. Fleishman, I. Drori, and D. Cohen-Or, “Bilateral mesh denoising,” in ACM SIGGRAPH 2003 Papers, San Diego California: ACM, Jul. 2003, pp. 950–953. doi: 10.1145/1201775.882368.
[4] J. Digne and C. de Franchis, “The Bilateral Filter for Point Clouds,” Image Processing On Line, vol. 7, pp. 278–287, Oct. 2017, doi: 10.5201/ipol.2017.179.
[5] H. Zhang, L. Zhu, X. Cai, and L. Dong, “Noise removal algorithm based on point cloud classification,” in 2022 International Seminar on Computer Science and Engineering Technology (SCSET), Indianapolis, IN, USA: IEEE, Jan. 2022, pp. 93–96. doi: 10.1109/SCSET55041.2022.00030.
[6] C. Lv and M. Li, “Point Cloud Denoising Algorithm Based on Noise Classification,” in 2020 International Conference on Culture-oriented Science & Technology (ICCST), Beijing, China: IEEE, Oct. 2020, pp. 123–127. doi: 10.1109/ICCST50977.2020.00029.
[7] K. Cao and P. Cosman, “Denoising and Inpainting for Point Clouds Compressed by V-PCC,” IEEE Access, vol. 9, pp. 107688–107700, 2021, doi: 10.1109/ACCESS.2021.3102029.
[8] Y. Ren, T. Li, J. Xu, W. Hong, Y. Zheng, and B. Fu, “Overall Filtering Algorithm for Multiscale Noise Removal From Point Cloud Data,” IEEE Access, vol. 9, pp. 110723–110734, 2021, doi: 10.1109/ACCESS.2021.3097185.
[9] R. Q. Charles, H. Su, M. Kaichun, and L. J. Guibas, “PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation,” in 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Honolulu, HI: IEEE, Jul. 2017, pp. 77–85. doi: 10.1109/CVPR.2017.16.
[10] W. Hou, D. Li, C. Xu, H. Zhang, and T. Li, “An Advanced k Nearest Neighbor Classification Algorithm Based on KD-tree,” 2018.
[11] H. Hoppe, T. DeRose, T. Duchamp, J. McDonald, and W. Stuetzle, “Surface reconstruction from unorganized points”.
[12] Leiou Wang, “A new fast median filtering algorithm based on FPGA,” in 2013 IEEE 10th International Conference on ASIC, Shenzhen, China: IEEE, Oct. 2013, pp. 1–4. doi: 10.1109/ASICON.2013.6812071.
[13] Booth, A. D. (1951). A signed binary multiplication technique. The Quarterly Journal of Mechanics and Applied Mathematics, 4(2), 236-240.(BOOTH)
[14] , A. Singh, and V. Niranjan, “Efficient FIR Filter Design using Booth Multiplier for VLSI Applications,” in 2018 International Conference on Computing, Power and Communication Technologies (GUCON), Sep. 2018, pp. 581–584. doi: 10.1109/GUCON.2018.8674998.
[15] Kumar, M. S., Inthiyaz, S., Babu, M. A., Teja, B., Ahmed, S. J., Harika, T. S., & Bhaskar, K. (2020). Two speed radix-2 booth multiplier using Verilog. Journal of Critical Reviews, 7(4), 668-671.
[16] J. Wei, A. Kuwana, H. Kobayashi, en K. Kubo, “Divide and Conquer: Floating-Point Exponential Calculation Based on Taylor-Series Expansion”, in 2021 IEEE 14th International Conference on ASIC (ASICON), Okt 2021, bll 1–4. doi: 10.1109/ASICON52560.2021.9620253.
[17] A. Javaheri, C. Brites, F. Pereira, and J. Ascenso, “Point Cloud Rendering After Coding: Impacts on Subjective and Objective Quality,” IEEE Trans. Multimedia, vol. 23, pp. 4049–4064, 2021, doi: 10.1109/TMM.2020.3037481.