A method and device for trajectory pattern recognition based on dynamic time warping

Abstract

The present invention proposes a method and device for identifying a trajectory pattern based on dynamic time warping. The method includes: uniformly resampling a given trajectory point to obtain a resampling point, so that the distance along the trajectory between every two consecutive points is equal; moving the resampling point by a fixed vector, so that the center of gravity of the moved point is at Origin; rotate the moved point in the specified direction by the specified angle value, where the specified angle value is the argument value of the first point; scale the rotated point so that the scaled point is within the specified size of the square; will process The resulting trajectory data is compared with each trajectory in the template library by the DTW algorithm, and the template with the smallest DTW distance is selected as its trajectory style. The invention can realize the pattern recognition of the two-dimensional trajectory.

Description

technical field [0001] The invention belongs to the field of pattern recognition, and in particular relates to a trajectory pattern recognition method and device based on dynamic time regularization, which are used for pattern recognition of various trajectories. Background technique [0002] Whether it is the pattern discrimination of the enemy's movement trajectory on the battlefield or the recognition of dynamic gestures, it can be classified as a problem of trajectory pattern recognition. The traditional dynamic time warping (Dynamic Time Warping) algorithm is often used in the pattern recognition of speech signals. When applied in trajectory recognition, since the data changes from one-dimensional to two-dimensional, the data preprocessing process is lacking. The initial position and velocity of the trajectories become sensitive, which makes the algorithm application difficult and the pattern recognition of the trajectory pattern inaccurate. SUMMARY OF THE INVENTION ...

AI Technical Summary

Problems solved by technology

The traditional Dynamic Time Warping (Dynamic Time Warping) algorithm is often used in the pattern recognition of speech signals. When it is applied in trajectory recognition, because the data changes from one-dimensional to two-dimensional, there is no preprocessing process for the data. The initial position and velocity become sensitive, which makes it difficult to apply the algorithm, and the pattern recognition of the trajectory pattern is inaccurate

Images