A self-tuning line feature extraction method for robot SLAM in structured environment

Aiming for the least squares method, all the points in the region are used as the features to be extracted. There is a problem that the computational complexity is large and it is easy influenced by edge interference information, which leads to inaccurate feature extraction. The self-tuning feature extraction method fusing mean clustering and K-L transform is proposed. The principal component analysis method is used to ensure that the extracted line features are consistent with the distribution of the main data points. The self-tuning feature extraction method can remove edge interference information through the clustering center calculation, parallel area division and other operations. It improves the accuracy of line feature extraction. Compared to the least squares method, K-L transform is faster and has better real-time performance.