Human body motion capturing data retrieval method based on bionic pattern recognition

Human body motion capturing data retrieval method based on bionic pattern recognition

  • CN 101,452,488 A
  • Filed: 10/11/2008
  • Published: 06/10/2009
  • Est. Priority Date: 10/11/2008
  • Status: Active Application
First Claim
Patent Images

1. , a kind of human body motion capturing data retrieval method based on bionical pattern-recognition, comprise will based on the human body motion capturing data retrieval system of C++ programming language and OpenGL shape library exploitation pack into computing machine and to the human body motion capture database make up, initialized step, it is characterized in that, further comprising the steps of:

  • (1), the search condition in this method be one group with the type games segment, choose key poses and with digital number (1,2,) serve as a mark, and be called retrieve sequence, for example certain " is walked " and obtains retrieve sequence [1 after the motion retrieval condition is chosen by key poses, 2,3,4];

    (2), at human body movement data (comprising motion retrieval database and search condition), with reference to BVH exercise data standard format, be that feature is carried out feature extraction, and the structural attitude vector with the rotation CHANNEL in the BVH standard format;

    (3), based on bionical pattern recognition theory and the application characteristic in the human body motion capture data thereof, proper vector at search condition, structure " super sausage " human motion feature overlay area in the higher dimensional space of correspondence, the dimension of this higher dimensional space equates with the proper vector dimension;

    (4), the shuttling movement database, judge the key poses classification in the search condition of all frame institute subordinates in the storehouse, and with digital number (1,2 ...) serve as a mark, the frame that is not subordinated to the key poses classification in any search condition is with sequence number 0 mark;

    (5), the definition sequence coverage, this sequence be in the motion database dependency marker of all frames in some motions (1,2 ...) arrangement, the quantity that the quantity of sequence coverage equals to move in the database;

    (6), estimate the similarity of sequence coverage and retrieve sequence, at first sequence coverage is simplified, short-cut method is for deleting 0 element in the sequence, adjacent identical element in the collating sequence, for example;

    sequence [1,1,0,2,0,0,3,3,0,4,4,4,0] be reduced to [1,2,3,4], the sequence coverage after the simplification is called the simplification sequence coverage, and the similarity evaluation standard of sequence coverage and retrieve sequence is for to judge whether retrieve sequence is the subsequence of sequence coverage, if, think similarly, add in the result for retrieval, otherwise think dissimilar;

View all claims
    ×
    ×

    Thank you for your feedback

    ×
    ×