CN102402289B - Mouse recognition method for gesture based on machine vision - Google Patents

Abstract

The invention discloses a mouse recognition method for gesture based on machine vision, comprising the following steps: (1) establishing an active shape model for gestures; (2) off-line training for gestures and getting a gesture features classifier; (3) collecting images; (4) extracting partial binary pattern characteristics for the images, searching the target gesture among the images through the gesture features classifier from step (2), then step (5); (5) positioning on the fingertips; (6) mapping of mouse. The invention is a natural and intuitive way of human-computer interaction, and has advantages of no need to carry other auxiliary equipments, finishing the mouse operation through natural hand and finger movement, small influence by light and background, and freely operation right switch when multi-users are operating the computer.

Description

A kind of mouse recognition method for gesture based on machine vision

Technical field

The present invention relates to machine vision technique and human-computer interaction technology, particularly a kind of mouse recognition method for gesture based on machine vision.

Background technology

Along with the development of computer technology, the interacting activity of people and computing machine becomes an important component part of mankind's schedule life.Mouse is mainly used in the man-machine interaction of graphical interfaces, but in reciprocal process, staff must be held a kind of mouse apparatus, in a plane, move, this mode of operation all has some limitations at aspects such as the naturality of man-machine interaction and friendly, and therefore research meets the development trend that man-machine interaction mode that the mankind are accustomed to naturally becomes field of human-computer interaction.

Human-computer interaction technology based on gesture mainly changes by identification user's gesture, the operation that judgement user need to carry out.It is divided into based on receipt gloves with based on machine vision two classes.Method based on data glove, requires user to wear a kind of sensor of similar gloves, thereby by computing machine, obtains position and the movable information of hand, and this method exists the shortcomings such as burdensome, dumb.The existing method based on machine, mostly adopts the method that the colour of skin is extracted to come dividing gesture region, and this method is subject to illumination effect large, and conventionally require the region of video acquisition be confined to hand region among a small circle in, this has limited the scope of activities of staff.

Summary of the invention

In order to overcome the deficiencies in the prior art, the object of the present invention is to provide a kind of natural, mouse recognition method for gesture based on machine vision intuitively.

Object of the present invention is achieved through the following technical solutions:

A mouse recognition method for gesture based on machine vision, comprises the following steps:

(1) set up the moving shape model of gesture: the profile of the gesture of first the five fingers being opened carries out the basic gesture modeling of B batten; Then on the basis of basic gesture model, set up the moving shape model of the gesture of B batten, shown in the following formula of its state space:

χ＝(x，y，θ，s，λ，θ ₁，l ₁，θ ₂，l ₂，θ ₃，l ₃，θ ₄，l ₄，θ ₅，θ ₆)

Wherein, x, y is respectively horizontal ordinate and the ordinate of the position of gesture; θ represents the gesture anglec of rotation planar; S represents the size of gesture; θ ₁, l ₁represent that respectively little finger is around angle and the little finger length of its root rotation; θ ₂, l ₂represent respectively nameless angle and nameless length around its root rotation; θ ₂, l ₃represent that respectively middle finger is around angle and the middle finger length of its root rotation; θ ₄, l ₄represent that respectively forefinger is around angle and the forefinger length of its root rotation; θ ₆represent that thumb is around the angle of its root rotation; θ ₅represent that thumb is around the angle of its joint, middle part rotation;

(2) gesture is carried out to off-line training, obtain gesture feature sorter;

(3) gather image;

(4) extract local binary pattern (LBP) feature of image, the gesture feature sorter obtaining by step (2) is searched for target gesture in image; When searching target gesture, carry out step (5);

(5) finger tip location:

(5-1) adopt the moving shape model of the gesture in step (1) to carry out evaluation fitting to the target gesture searching, obtain the state space initial value χ of target gesture ₀, make the initial value p of posterior probability density ₀(χ ₀| Z ₀)=1;

(5-2) target gesture is observed, utilized particle filter tracking algorithm to carry out iteration renewal to the state of target gesture, described iteration renewal process is specific as follows:

The observed quantity that defines t-1 two field picture is Z _t-1, the state space of target gesture is χ during t-1 frame _t-1posterior probability be p _t-1(χ _t-1| Z _t-1);

The optimal profile state of target gesture during according to t-1 frame the optimal profile state of target gesture during to t frame estimation probability be: p _t(χ _t| χ _t-1); Order

During t frame, the likelihood probability of the profile state of target gesture is p _t(Z _t| χ _t);

While obtaining t frame, the state space of target gesture is χ _tposterior probability be:

p _t(χ _t|Z _t)＝p _t(Z _t|χ _t)*∫p _t(χ _t|χ _t-1)*p _t-1(χ _t-1|Z _t-1)dχ _t-1/p _t(Z _t)；

The posterior probability density p of target gesture during according to t frame _t(χ _t| Z _t) the optimal profile state of target gesture while estimating t frame

(5-3) when t frame, in the optimal profile of target gesture, indicate the coordinate A of forefinger finger tip and the coordinate B of thumb finger tip;

(6) mouse mappings:

Build right-angle triangle ABC: cross A point and do vertical line, cross B point and do horizontal line, intersection point is C;

If line segment AB and CB angle are α, the angle of line segment BA and CA is β, and α is less than threshold value formula α _tbe mapped as left mouse button click action, β is less than threshold value beta _tbe mapped as and click action by mouse right button;

The mid point M of line taking section AB is mapped as the relative coordinate of mouse; If forefinger fingertip location is A in previous frame image ^*, the coordinate B of thumb finger tip ^*, line segment A ^*b ^*mid point be M ^*, directed line segment M*M is mapped as the relatively move vector of mouse on screen.

Step (2) is described carries out off-line training to gesture, obtains gesture feature sorter, is specially:

Use 600 open one's eyes wide mark gesture as positive sample and 1200 non-target gestures as negative sample; The opencv_traincascade program that use is increased income to be provided in the OpenCV of image processing class storehouse is trained.

The described gesture feature sorter obtaining by step (2) of step (4) is searched for target gesture in image, is specially:

The subwindow of intercepting entire image all 30 * 30, each subwindow, successively by 20 grades of gesture feature sorters, is eliminated non-target gesture subwindow step by step, and the subwindow by all 20 grades of gesture feature sorters is defined as target gesture; If do not find target gesture in this layer of search, by subwindow with 1.3 times of amplifications, again by gesture feature sorter, detect.

In the described search target of step (2) gesture process, if previous frame image has searched target gesture, the hunting zone of current frame image is reduced into region around, target gesture region in previous frame image graph picture, is specially:

If the center of the target gesture region searching in previous frame image is some O, some O is d to target gesture region left margin distance ₁, some O is d to right margin distance ₂, some O is d to coboundary distance ₃, some O is d to lower boundary distance ₄, new region of search is centered by an O, makees rectangular search frame, some O is 2*d to rectangular search frame left margin distance ₁, some O is 2*d to rectangular search frame right margin distance ₂, some O is 2*d to rectangular search upper frame edge circle distance ₃, some O is 2*d to rectangular search frame lower boundary distance ₄.

Step (5-2) is described to be observed target gesture, specifically by adaptive skin color segmentation method, observes; Described adaptive skin color segmentation method comprises the following steps:

(5-2-1) set up TSL space complexion model;

(5-2-2) carry out Complexion filter;

(5-2-3) bianry image obtaining after Complexion filter is carried out to dilation operation.

The described TSL of foundation of step (5-2-1) space complexion model is specially:

By following formula, by RGB color space conversion, be TSL color space:

$\left\{\begin{array}{c}T=\frac{1}{2\mathrm{\π}}{\tan }^{-1}\left(\frac{r^{\′}}{g^{\′}}\right)+0.5\\ S=\sqrt{\frac{9}{5}(r^{\′2}+g^{\′2})}\\ L=0.299*R+0.587*G+0.114*B\\ \end{array}\right.$

Wherein $r^{\′}=(r-\frac{1}{3}),$ $g^{\′}=(g-\frac{1}{3})$

$r=\frac{R}{R+G+B},$ $g=\frac{G}{R+G+B}$

R, G, B are respectively the RGB component under rgb color model; T, S, L are respectively the TSL component under TSL colour model.

Described in step (5-2-2), carry out Complexion filter, be specially:

The face of 500 images that comprise area of skin color and hand region are sampled, Mean Matrix E and the covariance matrix ∑ of the dimensional Gaussian distribution probability distribution parameter of T and S under estimation TSL model;

Each pixel is detected, if the mahalanobis distance of the C=(T, S) that the T of a pixel and S component form vector and mean vector E is lower than threshold value Threshold, think that this pixel belongs to area of skin color; Described mahalanobis distance d=(C-E) ^t∑ ^-1(C-E).

Described threshold value Threshold is determined by following process:

Distance according to Mean Matrix E and covariance matrix ∑ estimation C=(T, S) vector and mean vector E, obtains initial value;

According to following formula, calculate the degree of confidence confidence of each threshold value Threshold:

$\mathrm{Confidence}=\frac{\mathrm{PosSkin}}{\mathrm{MaskArea}}*2-\frac{\mathrm{NegSkin}}{\mathrm{BgArea}}$

Wherein PosSkin refers to the quantity of the pixel of the colour of skin in colour of skin template region, and MaskArea refers to the total area in colour of skin template region, and NegSkin refers to the quantity of skin pixel point in background template region, and BgArea refers to the area in background template region;

The threshold value Threshold of degree of confidence maximum is updated to the TSL complexion model parameter of current the best, and upgrades TSL space complexion model.

Compared with prior art, the present invention has the following advantages and technique effect:

1, the present invention is a kind of natural, man-machine interaction mode intuitively, and user does not need to carry other utility appliance, by natural hand and finger motion, completes mouse action;

2, particle filter tracking model under the comprehensive LBP feature of the present invention and ASM, carries out gesture zone location and profile and follows the tracks of, and is subject to illumination and background influence little;

3, the present invention has overcome conventional mouse and user has been operated to the restriction of degree of freedom, can make user before camera in 2m middle distance computing machine is controlled;

4, the present invention can be when multi-user operation computing machine, free blocked operation power.

Accompanying drawing explanation

Fig. 1 is the process flow diagram that the present invention is based on the mouse recognition method for gesture of machine vision.

Fig. 2 is the schematic diagram of the five fingers gesture of opening.

Fig. 3 is the schematic diagram of target gesture.

Fig. 4 is the schematic diagram in search target gesture process.

Fig. 5 is the template-setup schematic diagram in complexion model.

Fig. 6 builds the schematic diagram of right-angle triangle ABC in mouse mappings process.

Fig. 7 is that left mouse button is clicked schematic diagram.

Fig. 8 is for clicking schematic diagram by mouse right button.

Fig. 9 is the schematic diagram that mouse moves.

Embodiment

Below in conjunction with embodiment and accompanying drawing, the present invention is described in further detail, but embodiments of the present invention are not limited to this.

Embodiment

Mouse recognition method for gesture of the present invention is based on gesture mouse system, the gesture mouse system of the present embodiment is comprised of image capture module, image processing module and mouse event respond module, image capture module comprises camera, is responsible for Real-time Collection user's image and is transferred in image processing module.

Image processing module and mouse event respond module are completed by computing machine, and wherein image processing module moves various image processing algorithm real-time analysis user images, and the motion of user's hand and finger is converted to corresponding steering order.Mouse event respond module is accepted the steering order that image processing module sends, and is mouse response events by corresponding instruction transformation.

As shown in Figure 1, the mouse recognition method for gesture based on machine of the present embodiment, comprises the following steps:

(1) set up the moving shape model of gesture: the profile of the gesture of first the five fingers being opened (as shown in Figure 2) carries out the basic gesture modeling of B batten; Then on the basis of basic gesture model, set up the moving shape model of the gesture of B batten, shown in the following formula of its state space:

χ＝(x，y，θ，s，λ，θ ₁，l ₁，θ ₂，l ₂，θ ₃，l ₃，θ ₄，l ₄，θ ₅，θ ₆)

Wherein, x, y is respectively horizontal ordinate and the ordinate of the position of gesture; θ represents the gesture anglec of rotation planar; S represents the size of gesture; θ ₁, l ₁represent that respectively little finger is around angle and the little finger length of its root rotation; θ ₂, l ₂represent respectively nameless angle and nameless length around its root rotation; θ ₃, l ₃represent that respectively middle finger is around angle and the middle finger length of its root rotation; θ ₄, l ₄represent that respectively forefinger is around angle and the forefinger length of its root rotation; θ ₅, θ ₆represent that thumb is around the angle of its root rotation; θ ₆represent that thumb is around the angle of its root rotation; θ ₅represent that thumb is around the angle of its joint, middle part rotation;

(2) off-line training is carried out in staff region, obtains gesture feature sorter: use 600 open one's eyes wide mark gesture as positive sample and by 1200 non-target gestures as negative sample; The opencv traincascade program that use is increased income to be provided in the OpenCV of image processing class storehouse is trained.

(3) gather image;

(4) extract the partial binary mode characteristic of image, the gesture feature sorter obtaining by step (2) is searched for target gesture as shown in Figure 3 in image; When searching target gesture, carry out step (5);

The described gesture feature sorter obtaining by step (2) is searched for target gesture in image, is specially:

The subwindow of intercepting entire image all 30 * 30, each subwindow is successively by 20 grades of gesture feature sorters, eliminate step by step non-target gesture subwindow, only have the subwindow by all 20 grades of gesture feature sorters to be just defined as target gesture, if do not find target gesture in this layer of search, by subwindow with 1.3 times of amplifications, again by gesture feature sorter, detect.

In search procedure, in order to improve search efficiency, if previous frame image has searched target gesture, the hunting zone of current frame image is reduced into region around, target gesture region in previous frame image.As shown in Figure 4, suppose that the center, target gesture region (being the inside casing in Fig. 4) searching is some O in previous frame image, some O is d to the left margin distance of target gesture region ₁, some O is d to right margin distance ₂, some O is d to coboundary distance ₃, some O is at d to lower boundary distance ₄, new region of search is centered by an O, makees rectangular search frame (being the housing in Fig. 4), mid point O is 2*d to rectangular search frame left margin distance ₁, O is 2*d to rectangular search frame right margin distance ₂, O is 2*d to rectangular search upper frame edge circle distance ₃, O is 2*d to rectangular search frame lower boundary distance ₄.

(5) finger tip location:

(5-1) adopt the moving shape model of the gesture in step (1) to carry out evaluation fitting to the target gesture searching, obtain the state space initial value χ of target gesture ₀, make the initial value p of posterior probability density ₀(χ ₀| Z ₀)=1;

(5-2) by adaptive skin color segmentation method, target gesture is observed, utilized particle filter tracking algorithm to carry out iteration renewal to the state of target gesture;

Described adaptive skin color segmentation method comprises the following steps:

(5-2-1) set up TSL space complexion model:

First by following formula, by RGB color space conversion, be TSL color space:

$\left\{\begin{array}{c}T=\frac{1}{2\mathrm{\π}}{\tan }^{-1}\left(\frac{r^{\′}}{g^{\′}}\right)+0.5\\ S=\sqrt{\frac{9}{5}(r^{\′2}+g^{\′2})}\\ L=0.299*R+0.587*G+0.114*B\\ \end{array}\right.$

Wherein $r^{\′}=(r-\frac{1}{3}),$ $g^{\′}=(g-\frac{1}{3})$

$r=\frac{R}{R+G+B},$ $g=\frac{G}{R+G+B}$

R, G, B are respectively the RGB component under rgb color model; T, S, L are respectively the TSL component under TSL colour model;

(5-2-2) carry out Complexion filter: face and the hand region of the image by comprising area of skin color to 500 are sampled, estimate Mean Matrix E and the covariance matrix ∑ of the dimensional Gaussian distribution probability distribution parameter of T and S under TSL model;

Each pixel is detected, if the mahalanobis distance of the C=(T, S) that the T of a pixel and S component form vector and mean vector E is lower than threshold value Threshold, think that this pixel belongs to area of skin color; Described mahalanobis distance d=(C-E) ^t∑ ^-1(C-E);

Described threshold value Threshold is determined by following process:

Distance according to Mean Matrix E and covariance matrix ∑ estimation C=(T, S) vector and mean vector E, obtains initial value;

According to following formula, calculate the degree of confidence confidence of each threshold value Threshold:

$\mathrm{Confidence}=\frac{\mathrm{PosSkin}}{\mathrm{MaskArea}}*2-\frac{\mathrm{NegSkin}}{\mathrm{BgArea}}$

Template-setup in complexion model as shown in Figure 5.In above formula, PosSkin refers to the quantity of the pixel of the colour of skin in colour of skin template region (being gesture region in Fig. 5), MaskArea refers to the total area in colour of skin template region, NegSkin refers to the quantity of skin pixel point in background template region (being the black background in Fig. 5), and BgArea refers to the area in background template region.

The TSL complexion model parameter that the threshold value Threshold that gets degree of confidence maximum is current the best, and upgrade TSL space complexion model;

(5-2-3) bianry image obtaining after Complexion filter is carried out to dilation operation, the cavity causing to reduce Complexion filter.

Described iteration renewal process is specific as follows:

The observed quantity that defines t-1 two field picture is Z _t-1, the state space of target gesture is χ during t-1 frame _t-1posterior probability be p _t-1(χ _t-1| Z _t-1);

The optimal profile state of target gesture during according to t-1 frame during to t frame, the estimation probability of the optimal profile state of target gesture is: p _t(χ _t| χ _t-1); Order

During t frame, the likelihood probability of the profile state of target gesture is p _t(Z _t| χ _t);

While obtaining t frame, the state space of target gesture is χ _tposterior probability be:

p _t(χ _t|Z _t)＝p _t(Z _t|χ _t)*∫p _t(χ _t|χ _t-1)*p _t-1(χ _t-1|Z _t-1)dχ _t-1/p _t(Z _t)；

The posterior probability density p of target gesture during according to t frame _t(χ _t| Z _t) the optimal profile state of target gesture is while estimating t frame

(5-3) when t frame, in the optimal profile of target gesture, indicate the coordinate A of forefinger finger tip and the coordinate B of thumb finger tip;

(6) mouse mappings:

As shown in Figure 6, build right-angle triangle ABC: cross A point and do vertical line, cross B point and do horizontal line, intersection point is C, and establishing line segment AB and CB angle is α, and the angle of line segment BA and CA is β, and α is less than threshold alpha _tbe mapped as left mouse button click action, Fig. 7 is that left mouse button is clicked schematic diagram.β is less than threshold value beta _tbe mapped as and click action by mouse right button, Fig. 8 is for clicking schematic diagram by mouse right button.

The mid point M of line taking section AB is mapped as the relative coordinate of mouse, and establishing forefinger fingertip location in previous frame image is A ^*, the coordinate B of thumb finger tip ^*, line segment A ^*b ^*mid point be M ^*, directed line segment M*M is mapped as the relatively move vector of mouse on screen, and Fig. 9 is the schematic diagram that mouse moves.

Above-described embodiment is preferably embodiment of the present invention; but embodiments of the present invention are not limited by the examples; other any do not deviate from change, the modification done under Spirit Essence of the present invention and principle, substitutes, combination, simplify; all should be equivalent substitute mode, within being included in protection scope of the present invention.

Claims (8)

1. the mouse recognition method for gesture based on machine vision, is characterized in that, comprises the following steps:

(1) set up the moving shape model of gesture: the profile of the gesture of first the five fingers being opened carries out the basic gesture modeling of B batten; Then on the basis of basic gesture model, set up the moving shape model of the gesture of B batten, shown in the following formula of its state space:

$\mathrm{\χ}=(x,y,\mathrm{\θ},s,\mathrm{\λ},{\mathrm{\θ}}_1,l_1,{\mathrm{\θ}}_2,l_2,{\mathrm{\θ}}_3,l_3,{\mathrm{\θ}}_4,l_4,{\mathrm{\θ}}_5,{\mathrm{\θ}}_6)$

Wherein, x, y is respectively horizontal ordinate and the ordinate of the position of gesture; θ represents the gesture anglec of rotation planar; S represents the size of gesture; θ ₁, l ₁represent that respectively little finger is around angle and the little finger length of its root rotation; θ ₂, l ₂represent respectively nameless angle and nameless length around its root rotation; θ ₂, l ₂represent that respectively middle finger is around angle and the middle finger length of its root rotation; θ ₄, l ₄represent that respectively forefinger is around angle and the forefinger length of its root rotation; θ ₆represent that thumb is around the angle of its root rotation; θ ₅represent that thumb is around the angle of its joint, middle part rotation;

(2) gesture is carried out to off-line training, obtain gesture feature sorter;

(3) gather image;

(4) extract the partial binary mode characteristic of image, the gesture feature sorter obtaining by step (2) is searched for target gesture in image; When searching target gesture, carry out step (5);

(5) finger tip location:

(5-1) adopt the moving shape model of the gesture in step (1) to carry out evaluation fitting to the target gesture searching, obtain the state space initial value χ of target gesture ₀, make the initial value p of posterior probability density ₀(χ 0|Z ₀)=1;

(5-2) target gesture is observed, utilized particle filter tracking algorithm to carry out iteration renewal to the state of target gesture, described iteration renewal process is specific as follows:

The observed quantity that defines t-1 two field picture is Z _t-1, the state space of target gesture is χ during t-1 frame _t-1posterior probability be p _τ-1(χ _t-1| Z _τ-1);

The optimal profile state of target gesture during according to t-1 frame the optimal profile state of target gesture during to t frame estimation probability be: p _t(χ _t| χ _t-1); Order

During t frame, the likelihood probability of the profile state of target gesture is p _τ(z _τ| χ _τ);

During t frame, the state space of target gesture is χ _tposterior probability be:

p _t(χ _t｜z _t）=p _t（Ｚ _t｜χt:)*∫p _t(χ _t｜χ _t-1)*p _t-：（χ _t-1｜Ｚ _t-1:)dχ _t-1／p _t(Z _t)；

The posterior probability density p of target gesture during according to t frame _τ(χ _τ| while Z) estimating t frame, the optimal profile state of target gesture is

(5-3) when t frame, in the optimal profile of target gesture, indicate the coordinate A of forefinger finger tip and the coordinate B of thumb finger tip;

(6) mouse mappings:

Build right-angle triangle ABC: cross A point and do vertical line, cross B point and do horizontal line, intersection point is C;

If line segment AB and CB angle are α, the angle of line segment BA and CA is β, and α is less than threshold alpha _tbe mapped as left mouse button click action, β is less than threshold value beta _tbe mapped as and click action by mouse right button;

The mid point M of line taking section AB is mapped as the relative coordinate of mouse; If forefinger fingertip location is A in previous frame image ^*, the coordinate B of thumb finger tip ^*, line segment A ^*b ^*mid point be M*, directed line segment M*M is mapped as the relatively move vector of mouse on screen.

2. the mouse recognition method for gesture based on machine vision according to claim 1, is characterized in that, step (2) is described carries out off-line training to gesture, obtains gesture feature sorter, is specially:

Use 600 open one's eyes wide mark gesture as positive sample and 1200 non-target gestures as negative sample; The opencv_traincascade program that use is increased income to be provided in the OpenCV of image processing class storehouse is trained.

According to claim 1 based on machine depending on _ feel to it is characterized in that the in hand ca gesture sca mouse de mark recognition methods of tra, the described gesture feature sorter obtaining by step (2) of step (4) is searched for target gesture in image, is specially:

The subwindow of intercepting entire image all 30 * 30, each subwindow, successively by 20 grades of gesture feature sorters, is eliminated non-target gesture subwindow step by step, and the subwindow by all 20 grades of gesture feature sorters is defined as target gesture; If do not find target gesture in this layer of search, by subwindow with 1.3 times of amplifications, again by gesture feature sorter, detect.

4. the mouse recognition method for gesture based on machine vision according to claim 3, it is characterized in that, in described search target gesture process, if previous frame image has searched target gesture, the hunting zone of current frame image is reduced into region around, target gesture region in previous frame image, is specially:

If the center of the target gesture region searching in previous frame image is some O, some O is d to target gesture region left margin distance ₁, some O is d to right margin distance ₂, some O is d to coboundary distance ₃, some O is d to lower boundary distance ₄, new region of search is centered by an O, makees rectangular search frame, some O is 2*d to rectangular search frame left margin distance ₁, some O is 2*d to rectangular search frame right margin distance ₂, some O is 2*d to rectangular search upper frame edge circle distance ₃, some O is 2*d to rectangular search frame lower boundary distance ₄.

5. the mouse recognition method for gesture based on machine vision according to claim 1, is characterized in that, step (5-2) is described to be observed target gesture, specifically by adaptive skin color segmentation method, observes; Described adaptive skin color segmentation method comprises the following steps:

(5-2-1) set up TSL space complexion model;

(5-2-2) carry out Complexion filter;

(5-2-3) bianry image obtaining after Complexion filter is carried out to dilation operation.

6. the mouse recognition method for gesture based on machine vision according to claim 5, is characterized in that, the described TSL of foundation of step (5-2-1) space complexion model is specially:

By following formula, by RGB color space conversion, be TSL color space:

$\left\{\begin{array}{c}T=\frac{1}{2\mathrm{\π}}{\tan }^{-1}\left(\frac{r^{\′}}{g^{\′}}\right)+0.5\\ S=\sqrt{\frac{9}{5}(r^{\′2}+g^{\′2})}\\ L=0.299*R+0.587*G+0.114*B\end{array}\right.$

Wherein $r^{\′}=(r-\frac{1}{3}),g^{\′}=(g-\frac{1}{3})$

$r=\frac{R}{R+G+B},g=\frac{G}{R+G+B}$

R, G, B are respectively the RGB component under rgb color model; T, S, L are respectively the TSL component under TSL colour model.

7. the mouse recognition method for gesture based on machine vision according to claim 6, is characterized in that, carries out Complexion filter described in step (5-2-2), is specially:

The face of 500 images that comprise area of skin color and hand region are sampled, Mean Matrix E and the covariance matrix ∑ of the dimensional Gaussian distribution probability distribution parameter of T and S under estimation TSL model;

Each pixel is detected, if the mahalanobis distance of the C=(T, S) that the T of a pixel and S component form vector and mean vector E is lower than threshold value Threshold, think that this pixel belongs to area of skin color; Described Ma Shi is huge from d=(C-E) ^t∑ ^-1(C-E).

8. the mouse recognition method for gesture based on machine vision according to claim 7, is characterized in that, described threshold value Threshold is determined by following process:

Distance according to Mean Matrix E and covariance matrix ∑ estimation C=(T, S) vector and mean vector E, obtains initial value;

According to following formula, calculate the degree of confidence confidence of each threshold value Threshold:

$\mathrm{Confldence}=\frac{\mathrm{PosSkin}}{\mathrm{MaskArea}}*2-\frac{\mathrm{NegSkin}}{\mathrm{BgArea}}$

Wherein PosSkin refers to the quantity of the pixel of the colour of skin in colour of skin template region, and MaskArea refers to the total area in colour of skin template region, and NegSkin refers to the quantity of skin pixel point in background template region, and BgArea refers to the area in background template region;

The threshold value Threshold of degree of confidence maximum is updated to the TSL complexion model parameter of current the best, and upgrades TSL space complexion model.