Finger movement detection and positioning method based on space-time filtering and joint space Kmeans

Abstract

The invention discloses a finger movement detection and positioning method based on space-time filtering and joint space Kmeans. The finger movement detection and positioning method based on space-time filtering and joint space Kmeans includes the steps: attaching 10 different colors (except for black and white) of labels to the fingers of a player, shooting the video that the fingers of the player play a keyboard musical instrument; for the input video frames, using a space-time filtering method to detect the finger movement target, and feeding back the spatial domain filtering result to guide updating of the dynamic background; and using the joint space Kmeans to perform finger movement target positioning, and by means of combination of the statistical characteristics of R, G and B histograms to decide the cluster number and the initialization clustering center adaptively, so as to realize the fingering identification and recording function with low calculating complexity, high convergence speed, high positioning accuracy, and high real time performance.

Description

technical field [0001] The invention relates to the technical fields of visual monitoring, digital image processing, etc., in particular to a finger motion detection and positioning method based on spatio-temporal filtering and joint space Kmeans. Background technique [0002] Proper fingering for a piano (or other keyboard instrument) player is critical to dexterity in playing and interpreting music. Good fingering can reflect the performer's understanding and interpretation of the composer's style characteristics and the content of the work, and at the same time save energy and time and improve performance efficiency. Although there are general rules in playing fingering, the non-fixedness of fingering in different pieces makes it more difficult for beginners to practice fingering and to imitate the fingering of music masters. Manually recording fingerings not only requires high musical accomplishment, but also time-consuming and labor-intensive. Therefore, it is an inev...

AI Technical Summary

Problems solved by technology

Edge detection replaces simplified positioning information with precise target outline representation, but edge detection will lose a lot of information when the fingering is complex or the labels of two or more fingers overlap, and even two moving targets are judged as one

And because the edge detection can only locate but not classify, the finger cannot be correctly matched with the detected outline

At the same time, edge detection is greatly affected by the background, without filtering function, noise points will also be detected and interfere with finger positioning

[0009] Therefore, in view of various problems in the above-mentioned moving target detection and positioning method in the application scenario of playing fingering recognition, such as missed detection of low-speed moving targets, difficulty in classification, fingers cannot be correctly matched with positioning, and serious noise interference. And joint space Kmeans finger motion detection and localization method, by analyzing the video of the player's fingers playing the piano (or other keyboard instruments) to realize fingering recognition

Images