研究目的
To present a method based on Sparse code shrinkage algorithm to reduce noise effect in final reconstructed 3D model from underwater laser scanning images.
研究成果
SCS effectively separates the signal from the noise and removes most of the noise by setting small components to zero, showing superior performance in denoising underwater laser scanning images without losing too many features. However, its time-consuming nature suggests a need for future research to speed up the process.
研究不足
SCS is a time-consuming process, particularly its sub-process ICA estimation, which depends on the image size and ICA transform window size. It may take up to ten minutes for processing, indicating a need for optimization in future research.
1：Experimental Design and Method Selection:
The study employs the Sparse Code Shrinkage (SCS) algorithm for denoising underwater laser scanning images, comparing its performance with Gaussian, median, and Wiener filters.
2：Sample Selection and Data Sources:
Natural and real underwater scanned images from a 3D scanner are used, with images corrupted by different levels of Gaussian noise to simulate real conditions.
3：List of Experimental Equipment and Materials:
A linear stage with 1mm moving step and a flat wall placed vertically to the scanner are used.
4：Experimental Procedures and Operational Workflow:
Images are denoised using Gaussian, median, Wiener filters, and SCS, with parameters adjusted for best results. The process includes generating noise-free data, estimating sparse coding transformation, applying shrinkage operators, and transforming back to original variables.
5：Data Analysis Methods:
Orthogonal linear regression fitting is applied to estimate roughness/smoothness of 3D models, with measurements taken from specific features to evaluate accuracy.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
