A kind of method by the action of computer vision system identification human body limb
Technical field
The present invention relates to a kind of method by the action of computer vision system identification human body limb.
Background technology
Traditional computer vision system based on graphical analysis need be analyzed RGB figure, and its data volume is big and speed is not good, and the algorithm that adopts is affected by environment bigger around feature or the contrast of bitmap fuzzy algorithm, comparatively unstable.
Summary of the invention
The objective of the invention is to overcome the defective that prior art exists, a kind of method by the action of computer vision system identification human body limb is provided, its computational data amount is little, computing velocity fast and affected by environment little, stable.
The technical solution adopted for the present invention to solve the technical problems is: a kind of method by the action of computer vision system identification human body limb, and step is as follows: a, catch image by camera, and a picture that captures is kept in the computing machine; B, computing machine carry out gray scale according to color psychology formula: brightness=R*0.299+G*0.587+B*0.114 to the picture of preserving to be handled; C, the gray scale picture is carried out the border processing; D, picture is divided into a plurality of independent gradation zones according to boundary value; E, generate corresponding figure abstract object, and this figure abstract object is kept in the computing machine with the border singular point and the smooth arc degree of each isolated area; F, the position of the figure abstract object of figure abstract object and former frame is compared the movement locus of judging moving object in the picture.
Further, in step c, the gray scale picture is carried out border Hua Chu according to even gradual change formula: gradual change=A+ (B-A)/Step*N, get suitable color range step and territory N and calculate certain some color gradient region of rejection on every side as threshold values, drop on a little in the gradual change region of rejection as around this point, then this point is called singular point, and non-isolated singular point gets up to become the border continuously.
Beneficial effect of the present invention: the invention solves the defective that exists in the background technology, adopt the feasible process that will change into the process that RGB handles originally of the psychologic theory of color to brightness processed, its data volume is little, improved processing speed greatly, therefore the speed of judging for graphic limit also significantly improve, the motion capture process no longer relies on simple pixel comparison, but the element in the figure is carried out mathematical abstractions, make the element among the figure to discern, can inquire about, affected by environment little, stable.
Embodiment
A kind of method by the action of computer vision system identification human body limb, step is as follows: a, catch image by camera, and a picture that captures is kept in the computing machine; B, computing machine carry out gray scale according to color psychology formula: brightness=R*0.299+G*0.587+B*0.114 to the picture of preserving to be handled; C, basis evenly gradual change formula: gradual change=A+ (B-A)/Step*N are carried out border Hua Chu to the gray scale picture, get suitable color range step and territory N and calculate certain some color gradient region of rejection on every side as threshold values, drop on a little in the gradual change region of rejection as around this point, then this point is called singular point, and non-isolated singular point gets up to become the border continuously; D, picture is divided into a plurality of independent gradation zones according to boundary value; E, generate corresponding figure abstract object, and this figure abstract object is kept in the computing machine with the border singular point and the smooth arc degree of each isolated area; F, the position of the figure abstract object of figure abstract object and former frame is compared the movement locus of judging moving object in the picture.
Adopting said method is caught the limb action of the preceding human body of camera, define the limbs dummy command of thinking correspondence according to different movement track and limbs change of shape, be associated with the operation of other electronic equipments again, can under the situation that does not contact any object, carry out virtualization operations electronic equipment.
Should be appreciated that specific embodiment described above only is used to explain the present invention, and be not used in qualification the present invention.Conspicuous variation of being extended out by spirit of the present invention or change still are among protection scope of the present invention.