Method for extracting local feature points of partially-occluded object

The invention provides a method for extracting local feature points of a partially-occluded object. The technical scheme is that the method comprises the steps of firstly extracting the local feature points which are stable under the condition that most of feature points change based on a scale invariant feature transforming algorithm, and using the local feature points as coarse feature points; then performing confidence coefficient calculation of the coarse feature points in three aspects including gray level contrast ratio, gradient mean value amplitude ratio and high curvature and setting threshold values to screen the coarse feature points; and finally extracting local steady feature points from screened coarse feature points. The problem of low extracting and calculating speed of the local feature points of a large data volume of existing images is solved, and meanwhile the robustness of the local feature points is ensured.