This paper proposes a local feature point detection and matching method suitable for target tracking. It makes many improvements based on the Scale Invariant Feature Transform (SIFT) algorithm. Only local maxima are detected in the Gaussian difference scale space to improve the stability of the algorithm. The main direction and descriptor of the point of interest are determined based on the circular neighborhood statistical gradient direction histogram, avoiding the computational cost of image rotation. Finally, the ratio of the nearest neighbor to the next nearest neighbor is used to match the 96-dimensional descriptor. The proposed method not only effectively improves the matching accuracy, but also greatly improves the computing speed, which is suitable for occasions with high real-time requirements.
You Might Like
Recommended ContentMore
Open source project More
Popular Components
Searched by Users
Just Take a LookMore
Trending Downloads
Trending ArticlesMore