The 3D Point Cloud Registration Algorithm Based on Harris-DLFS

Three-dimensional model reconstruction is a pivotal technology in the realm of computer vision. Point cloud registration serves as its integral step, which decisively impacts efficiency and precision entire process. However, existing point algorithms often face issues. These include prolonged processing time, inadequate accuracy, poor robustness. To address these problems, this paper proposes novel algorithm based on corner detection (Harris) partition-based local feature statistics (DLFS). The main steps are follows: Firstly, Harris employed. This step crucial for extracting key points enhancing Secondly, DLFS method used to describe features each point, generating vectors. Subsequently, matching pairs filtered rigid distance constraints, an coarse performed using Random Sample Consensus (RANSAC) algorithm. Finally, Iterative Closest (ICP) applied fine registration. Experimental results demonstrated effectiveness method. It significantly improved robustness, computational efficiency. Therefore, it holds substantial value practical applications.