CN105335711A - Fingertip detection method in complex environment - Google Patents

Abstract

The invention provides a fingertip detection method in a complex environment. The fingertip detection method comprises the steps of: step 1, calculating dense light stream information corresponding to scene information, and reconstructing a skin color filter to obtain a hand region; step 2, and constructing models of the hand region in various gestures by adopting equal-area blocks, calculating a mass center of the hand region, calculating distances from all contour sampling points to the mass center and an average mass center distance, determining an extended mass center distance according to detected number of fingertips, drawing a circle by taking the mass center as a circle center and the extended mass center distance as the radius, removing contour points inside the circle and a wrist region with maximum number of continuous pixels on the circle, searching contour points with partial maximum mass center distance outside the circle, and marking the contour points as fingertips, and comparing the detected number of fingertips in this round with the detected number of fingertips in the last round to judge whether to continue the fingertip detection. The fingertip detection method is high in robustness, and can detect the fingertips correctly when the hand of a person moves in front of a camera freely in the complex environment, thereby increasing the accuracy and effectiveness of fingertip detection.

Description

Fingertip Detection under a kind of complex environment

Technical field

The present invention relates to image processing and analysis technical field, more particularly, relate to Fingertip Detection under a kind of complex environment.

Background technology

Traditional man-machine interactive system carries out the information interchange of the mankind and computing machine mainly through media such as button, mouse, keyboards, and these interactive modes exist that function is limited, take up room the shortcoming such as large, constrains the development of human-computer interaction technology mostly.In recent years, along with the development of computer science and artificial intelligence technology, man-machine interaction trends towards nature, intuitively mode gradually, wherein, man-machine interaction mode based on computer vision technique relies on the features such as it is easy to use, dirigibility is high, noise resisting ability is strong, obtains the concern of more and more researchist.

As the important component part of the man-machine interactive system based on computer vision technique, finger tip detects has very wide application prospect in the field such as gesture identification, virtual controlling.Early stage finger tip detection is marked by special pigment or is worn the modes such as LED from scene, detects finger tip, although this method is simple, uses very inconvenient, can only be used for some simple scene; Thereafter, the use of Wearable data glove is in the robustness that improve finger tip detection to a certain extent, but its convenience is poor, and Wearable data glove price general charged is higher simultaneously; Along with the development of camera technology, researchist brings into use some special cameras (as Kinect) to obtain special scene information, although finger tip Detection results is significantly improved, such camera popularization is not high, is difficult to promote.Meanwhile, have many researchists to propose the algorithm adopting common camera to carry out finger tip detection, but these algorithms are mostly fairly simple, cannot carry out staff detection under complex environment.In sum, current employing common camera capturing scenes information carries out the research tendency that finger tip detects still man-machine hand interactive system.

Still there is an a lot of difficult problem and not yet solve in current finger tip detection technique: (1) staff is non-rigid object, has very high degree of freedom, therefore cannot mate the finger under different situations by fixing template; (2) there is difference in the finger outward appearance of different people, accurately will detect finger shape and the attitude of different people, have larger difficulty; (3) even the finger of same person, also there is fine distinction, if cannot correctly identify these nuances, comparatively big error will be brought to testing result; (4) from scene, split hand region is the basis of carrying out finger tip detection, and when environment is comparatively complicated, the Accurate Segmentation of hand region has larger difficulty.

Summary of the invention

The object of the invention is to overcome shortcoming of the prior art with not enough, Fingertip Detection under a kind of complex environment is provided; This Fingertip Detection not only effectively can realize the segmentation carrying out hand region under complex environment, and this detection method strong robustness, also correctly finger tip be can detect when staff freely moves before camera under complex environment, thus degree of accuracy and the validity of finger tip detection improved.

In order to achieve the above object, the present invention is achieved by following technical proposals: Fingertip Detection under a kind of complex environment, for detecting the finger tip of hand; It is characterized in that: comprise following three steps:

The first step, capturing scenes information, by calculating the dense optical flow information corresponding with scene information and carrying out the scene image that binaryzation pre-service obtains containing hand region; Build Complexion filter device again, the scene image containing hand region is carried out to the segmentation of hand region, obtain hand region;

Second step, adopts homalographic block to build the model of each attitude hand region, according to the barycenter of the areal calculation hand region of hand region; Again hand region contour is sampled, calculate the distance of all configuration samplings point to barycenter, and calculate average barycenter distance; Then determine extended centroid distance according to the last round of finger tip number detected, and be the center of circle with barycenter, extended centroid is apart from carrying out picture circle for radius; Finally reject the point in circle and the circle wrist area that above contiguous pixels number is maximum, and in the outer point finding the maximum barycenter distance in local of circle, then this point is labeled as finger tip; This is taken turns and detects that the number of finger tip compares with the last round of finger tip number detected, if both are equal, then realize the detection of finger tip, otherwise this is taken turns detect that the number of finger tip is updated to the finger tip number detected, determine extended centroid distance and carry out picture circle, continuing the detection of finger tip.

In a first step, described capturing scenes information, by calculating the dense optical flow information corresponding with scene information and carrying out the scene image that binaryzation pre-service obtains containing hand region; Build Complexion filter device again, the scene image containing hand region carried out to the segmentation of hand region, obtain hand region and refer to: comprise the following steps:

(1.1) capturing scenes information, and calculate dense optical flow information corresponding with it;

(1.2) travel through Optic flow information, find maximum light stream value, and utilize this maximum light stream value, in X-axis and Y direction, Regularization is carried out to all light streams;

(1.3) according to the light stream after regularization, calculate tone and the saturation degree in the region changed along light stream direction, and mark different color values;

(1.4) set threshold value, according to threshold value, the light stream region of variation of color mark is converted into bianry image, in conjunction with logical operation and mathematical morphological operation, obtain the scene image containing hand region.

(1.5) according to human body complexion Clustering features, structure YCbCr Complexion filter device, eliminate redundancy color and monochrome information;

(1.6) after YCbCr Complexion filter device being filtered, sort descending is carried out to all profiles of the scene image containing hand region, find largest connected region as hand region, realize the segmentation scene image containing hand region being carried out to hand region.

In second step, described employing homalographic block builds the model of each attitude hand region, determines the area of hand region, to calculate the barycenter of hand region; Again hand region contour is sampled, calculate the distance of all configuration samplings point to barycenter, and calculate average barycenter distance; Then determine extended centroid distance according to the last round of finger tip number detected, and be the center of circle with barycenter, extended centroid is apart from carrying out picture circle for radius; Finally reject the point in circle and the circle wrist area that above contiguous pixels number is maximum, and in the outer point finding the maximum barycenter distance in local of circle, then this point is labeled as finger tip; This is taken turns and detects that the number of finger tip compares with the last round of finger tip number detected, if both are equal, then realize the detection of finger tip, otherwise this is taken turns and detects that the number of finger tip is updated to the finger tip number detected, determine extended centroid distance and carry out picture circle, the detection continuing finger tip refers to: comprise the following steps:

(2.1) build hand region display model with homalographic block, according to the different attitudes of hand region, obtain the quantity of homalographic block, determine the area of hand region;

(2.2) barycenter of hand region is calculated according to the area of hand region and the coordinate of each homalographic block;

(2.3) intensive sampling is carried out to hand region contour, calculate the distance of point to barycenter of all samplings, and calculate average barycenter distance;

(2.4) according to the finger tip number N determination extended centroid distance D detected _ext, and be the center of circle with barycenter, extended centroid is apart from D _extfor radius carries out picture circle; And reject the point be included in this circle; Wherein, the initial value of N is set to 0;

(2.5) add up the number of pixels of this circle through hand region, and determine that the region that the upper contiguous pixels number of circle is maximum is wrist area, and reject the point that this wrist area comprises;

(2.6) the outer remaining region of circle is found to the point of the maximum barycenter distance in local, and this point is labeled as finger tip;

(2.7) number N of finger tip will be detected ₁compare with the finger tip number N detected in step (2.4), if both are equal, then realize the detection of finger tip, obtain number and the position of finger tip; Otherwise will the number N of finger tip be detected ₁be updated to the finger tip number N detected in step (2.4), and return step (2.4), continue the detection of finger tip.

In step (2.2), the barycenter that the described area according to hand region calculates hand region with the coordinate of each homalographic block refers to: the barycenter (x being calculated hand region by formula (1) _g, y _g):

$x_g=\frac{s\underset{i=1}{\overset{n}{\Σ}}{x}_i}{s\underset{i=1}{\overset{n}{\Σ}}1}=\frac{\underset{i=1}{\overset{n}{\Σ}}{x}_i}{n},y_g=\frac{s\underset{i=1}{\overset{n}{\Σ}}{y}_i}{s\underset{i=1}{\overset{n}{\Σ}}1}=\frac{\underset{i=1}{\overset{n}{\Σ}}{y}_i}{n}---\left(1\right)$

Wherein, S is the area of hand region, x _i, y _ithat i-th homalographic block is at X-direction and Y direction coordinate respectively.

In step (2.3), the point of all samplings of described calculating to the distance of barycenter, and calculates average barycenter apart from referring to: the point (x calculating all samplings _{point_i}, y _{point_i}) to the Euclidean distance of barycenter, average barycenter distance formula (2) calculates:

$D_{avg}=\frac{\∫\∫\sqrt{{(x_{po\mathrm{int}\_i}-x_g)}^2+{(y_{po\mathrm{int}\_i}-y_g)}^2}dxdy}{\∫\∫dxdy}---\left(2\right)$

Wherein, (x _g, y _g) be the barycenter of hand region.

In step (2.4), the finger tip number N determination extended centroid that described basis has detected is apart from D _extrefer to: the described finger tip number N that detected and extended centroid are apart from D _extthe condition met is:

0≤N<3,D _ext＝1.5×D _avg

3≤N≤5,D _ext＝1.2×D _avg

If the finger tip number N >5 detected, then return the first step; Wherein, D _avgfor average barycenter distance.

In step (2.6), the described point outer remaining region of circle being found to the maximum barycenter distance in local, and this point is labeled as finger tip refers to: by traversal finger tip point, when the barycenter of 10 point continuous in point is apart from when increasing gradually, record current maximum, when barycenter is apart from the barycenter of continuous 10 point apart from when starting to reduce, using the current maximal value recorded as the maximum barycenter distance in local, the maximum barycenter in local is then labeled as finger tip apart from corresponding point.

Compared with prior art, tool of the present invention has the following advantages and beneficial effect:

1, be different from existing additive method, do not need to use special camera or specific installation when this method is implemented, also do not need to carry out special marking to staff, staff freely can move before camera.

Wherein, when carrying out hand region segmentation, by visual for the light stream region of variation of the video scene of catching, and by structure YCbCr Complexion filter device and the profile descending mode of screening, effectively realize the hand region segmentation under complex environment, at utmost avoid the interference of neighbourhood noise, the situation comprising a large amount of class area of skin color can be tackled in scene.

When carrying out finger tip to hand region and detecting, this method constructs hand region homalographic block models, calculates average barycenter apart from (D by the analytical algorithm of this model _avg) and extended centroid distance (D _ext), and with D _avgwith different D _extfor radius draws circle, carry out wrist set direction and point screening, eventually through the maximum barycenter in calculating local apart from finding fingertip location, flase drop measuring point in effective eliminating circle, thus effectively reject the interference that the noise information such as wrist, the tiny projection of hand detects finger tip, guarantee that finger tip point is positioned at outside circle, the finger tip realizing robust detects.In addition, because circle has rotational invariance, therefore when palm is rotated into different angles, the method still correctly can detect finger tip.

2, this Fingertip Detection not only effectively can realize the segmentation carrying out hand region under complex environment, and this detection method strong robustness, also correctly finger tip be can detect when staff freely moves before camera under complex environment, thus degree of accuracy and the validity of finger tip detection improved.

Accompanying drawing explanation

Fig. 1 is the FB(flow block) of Fingertip Detection of the present invention;

Fig. 2 is the method flow diagram that the first step carries out hand region segmentation from complex environment;

Fig. 3 is that second step carries out the method flow diagram of finger tip detection to the hand region split;

Fig. 4 is palm the five fingers hand region display model figure when opening;

Fig. 5 is palm hand region display model figure when presenting fist;

Fig. 6 be palm when presenting fist hand region be reduced to the schematic diagram of geometry appearance model;

When Fig. 7 is palm list finger tip, hand region is reduced to the schematic diagram of geometry appearance model;

Embodiment

Below in conjunction with accompanying drawing and embodiment, the present invention is described in further detail.

Embodiment

As shown in Figures 1 to 3, Fingertip Detection under complex environment of the present invention, for detecting the finger tip of hand; It is characterized in that: comprise following three steps:

The first step, capturing scenes information, by calculating the dense optical flow information corresponding with scene information and carrying out the scene image that binaryzation pre-service obtains containing hand region; Build Complexion filter device again, the scene image containing hand region is carried out to the segmentation of hand region, obtain hand region;

Second step, adopts homalographic block to build the model of each attitude hand region, according to the barycenter of the areal calculation hand region of hand region; Again hand region contour is sampled, calculate the distance of all configuration samplings point to barycenter, and calculate average barycenter distance; Then determine extended centroid distance according to the last round of finger tip number detected, and be the center of circle with barycenter, extended centroid is apart from carrying out picture circle for radius; Finally reject the point in circle and the circle wrist area that above contiguous pixels number is maximum, and in the outer point finding the maximum barycenter distance in local of circle, then this point is labeled as finger tip; This is taken turns and detects that the number of finger tip compares with the last round of finger tip number detected, if both are equal, then realize the detection of finger tip, otherwise this is taken turns detect that the number of finger tip is updated to the finger tip number detected, determine extended centroid distance and carry out picture circle, continuing the detection of finger tip.

In a first step, described capturing scenes information, by calculating the dense optical flow information corresponding with scene information and carrying out the scene image that binaryzation pre-service obtains containing hand region; Build Complexion filter device again, the scene image containing hand region carried out to the segmentation of hand region, obtain hand region and refer to: comprise the following steps:

(1.1) capturing scenes information, and calculate dense optical flow information corresponding with it; Although the calculated amount of dense optical flow is general all very large, but in the method, because the present invention calculates dense optical flow just in order to obtain hand candidate approximate region, therefore we use double-deck pyramid to carry out optical flow computation, and the search window (15x15) that setting is larger.

(1.2) travel through Optic flow information, find maximum light stream value, and utilize this maximum light stream value, in X-axis and Y direction, Regularization is carried out to all light streams.

(1.3) according to the light stream after regularization, calculate tone and the saturation degree in the region changed along light stream direction, and mark different color values.

(1.4) set threshold value, according to threshold value, the light stream region of variation of color mark is converted into bianry image, in conjunction with logical operation and mathematical morphological operation, obtain the scene image containing hand region.

(1.5) according to human body complexion Clustering features, structure YCbCr Complexion filter device, eliminate redundancy color and monochrome information; Generally, candidate's hand region image comprises RGB colouring information, it can be converted into YCbCr colouring information according to formula (3).

$\left[\begin{array}{c}Y\\ Cb\\ Cr\\ 1\end{array}\right]=\left[\begin{array}{cccc}0.2990& 0.5870& 0.1140& 0\\ -0.1687& -0.3313& 0.5000& 128\\ 0.5000& -0.4187& -0.0813& 128\\ 0& 0& 0& 1\end{array}\right]\left[\begin{array}{c}R\\ G\\ B\\ 1\end{array}\right]---\left(3\right)$

Meanwhile, the distribution due to human body complexion has obvious Clustering features, and in conjunction with a large amount of experimental results, the long and narrow colour band that formula (4) can be provided as complexion model, and constructs YCbCr Complexion filter device.

$\left\{\begin{array}{c}100\&le;Cb\&le;127\\ 128\&le;Cr\&le;170\end{array}\right.---\left(4\right)$

(1.6) after YCbCr Complexion filter device being filtered, sort descending is carried out to all profiles of the scene image containing hand region, find largest connected region as hand region, realize the segmentation scene image containing hand region being carried out to hand region.

In second step, described employing homalographic block builds the model of each attitude hand region, determines the area of hand region, to calculate the barycenter of hand region; Again hand region contour is sampled, calculate the distance of all configuration samplings point to barycenter, and calculate average barycenter distance; Then determine extended centroid distance according to the last round of finger tip number detected, and be the center of circle with barycenter, extended centroid is apart from carrying out picture circle for radius; Finally reject the point in circle and the circle wrist area that above contiguous pixels number is maximum, and in the outer point finding the maximum barycenter distance in local of circle, then this point is labeled as finger tip; This is taken turns and detects that the number of finger tip compares with the last round of finger tip number detected, if both are equal, then realize the detection of finger tip, otherwise this is taken turns and detects that the number of finger tip is updated to the finger tip number detected, determine extended centroid distance and carry out picture circle, the detection continuing finger tip refers to: comprise the following steps:

(2.1) build hand region display model with homalographic block, according to the different attitudes of hand region, obtain the quantity of homalographic block, determine the area of hand region; As shown in Figure 4, when the palm the five fingers open, in conjunction with hand Degrees of Freedom Model, hand region display model can be represented with 26 homalographic blocks; As shown in Figure 5, when palm presents fist, then available 16 homalographic blocks represent its display model.For the ease of calculating, for fist display model, this model can be reduced to geometry appearance model further, as shown in Figure 6.

(2.2) barycenter of hand region is calculated according to the area of hand region and the coordinate of each homalographic block; Barycenter (the x of hand region is calculated particular by formula (1) _g, y _g):

$x_g=\frac{s\underset{i=1}{\overset{n}{\&Sigma;}}{x}_i}{s\underset{i=1}{\overset{n}{\&Sigma;}}1}=\frac{\underset{i=1}{\overset{n}{\&Sigma;}}{x}_i}{n},y_g=\frac{s\underset{i=1}{\overset{n}{\&Sigma;}}{y}_i}{s\underset{i=1}{\overset{n}{\&Sigma;}}1}=\frac{\underset{i=1}{\overset{n}{\&Sigma;}}{y}_i}{n}---\left(1\right)$

Wherein, S is the area of hand region, x _i, y _ithat i-th homalographic block is at X-direction and Y direction coordinate respectively.

(2.3) intensive sampling is carried out to hand region contour, calculate the distance of point to barycenter of all samplings, and calculate average barycenter distance; Calculate the point (x of all samplings _{point_i}, y _{point_i}) to the Euclidean distance of barycenter, average barycenter distance formula (2) calculates:

$D_{avg}=\frac{\&Integral;\&Integral;\sqrt{{(x_{po\mathrm{int}\_i}-x_g)}^2+{(y_{po\mathrm{int}\_i}-y_g)}^2}dxdy}{\&Integral;\&Integral;dxdy}---\left(2\right)$

Wherein, (x _g, y _g) be the barycenter of hand region.

(2.4) according to the finger tip number N determination extended centroid distance D detected _ext, and be the center of circle with barycenter, extended centroid is apart from D _extfor radius carries out picture circle; And reject the point be included in this circle.Wherein, the initial value of N is set to 0; ; The finger tip number N detected and extended centroid are apart from D _extthe condition met is:

0≤N<3,D _ext＝1.5×D _avg

3≤N≤5,D _ext＝1.2×D _avg

If the finger tip number N >5 detected, then return the first step; Wherein, D _avgfor average barycenter distance.

Based on hand display model, when palm presents different finger number, the number of homalographic block is different, therefore draws circle with different extended centroid apart from for radius, effectively wrist area can be rejected the interference of tiny protruding point.In conjunction with hand outward appearance geometric model and formula (1), formula (2), when finger number is less than 3, extended centroid is apart from the average barycenter distance (D that should be 1.5 times _ext=1.5 × D _avg); When finger number is greater than 3 but is less than 5, extended centroid is apart from being 1.2 times of average barycenter distance (D _ext=1.2 × D _avg); When finger number is greater than 5, terminates finger tip and detect, do not do with extended centroid apart from the circle for radius.For the fist geometry appearance model shown in Fig. 6, barycenter is positioned at center, and draw circle using 1.5 times of average barycenter apart from as radius, this circle will comprise homalographic blocks all on fist, thus point all on fist all be rejected, and can not go out finger tip by flase drop.Single finger tip geometry appearance model as shown in Figure 7 again, circle is drawn apart from as radius using 1.5 times of average barycenter, except the finger tip of projection, other most of area block all will be included in extended centroid distance as in the circle of radius, thus effectively reject the noise spot of tiny projection, determined with extended centroid distance for the fingertip location outside the circle of radius by subsequent calculations.

(2.5) add up the number of pixels of this circle through hand region, and determine that the region that the upper contiguous pixels number of circle is maximum is wrist area, finger tip detection is not carried out in this direction, and rejects the point that this wrist area comprises.

(2.6) the outer remaining region of circle is found to the point of the maximum barycenter distance in local, and this point is labeled as finger tip.Step (2.6) is specifically such: by traversal finger tip point, when the barycenter of 10 point continuous in point is apart from when increasing gradually, record current maximum, when barycenter apart from continuous 10 point barycenter apart from start to reduce time, using the current maximal value recorded as the maximum barycenter distance in local, the maximum barycenter in local is then labeled as finger tip apart from corresponding point.

(2.7) number N of finger tip will be detected ₁compare with the finger tip number N detected in step (2.4), if both are equal, then realize the detection of finger tip, obtain number and the position of finger tip; Otherwise will the number N of finger tip be detected ₁be updated to the finger tip number N detected in step (2.4), and return step (2.4), continue the detection of finger tip.

Above-described embodiment is the present invention's preferably embodiment; but embodiments of the present invention are not restricted to the described embodiments; change, the modification done under other any does not deviate from Spirit Essence of the present invention and principle, substitute, combine, simplify; all should be the substitute mode of equivalence, be included within protection scope of the present invention.

Claims (7)

1. a Fingertip Detection under complex environment, for detecting the finger tip of hand; It is characterized in that: comprise following three steps:

The first step, capturing scenes information, by calculating the dense optical flow information corresponding with scene information and carrying out the scene image that binaryzation pre-service obtains containing hand region; Build Complexion filter device again, the scene image containing hand region is carried out to the segmentation of hand region, obtain hand region;

Second step, adopts homalographic block to build the model of each attitude hand region, according to the barycenter of the areal calculation hand region of hand region; Again hand region contour is sampled, calculate the distance of all configuration samplings point to barycenter, and calculate average barycenter distance; Then determine extended centroid distance according to the last round of finger tip number detected, and be the center of circle with barycenter, extended centroid is apart from carrying out picture circle for radius; Finally reject the point in circle and the circle wrist area that above contiguous pixels number is maximum, and in the outer point finding the maximum barycenter distance in local of circle, then this point is labeled as finger tip; This is taken turns and detects that the number of finger tip compares with the last round of finger tip number detected, if both are equal, then realize the detection of finger tip, otherwise this is taken turns detect that the number of finger tip is updated to the finger tip number detected, determine extended centroid distance and carry out picture circle, continuing the detection of finger tip.

2. Fingertip Detection under complex environment according to claim 1, it is characterized in that: in a first step, described capturing scenes information, by calculating the dense optical flow information corresponding with scene information and carrying out the scene image that binaryzation pre-service obtains containing hand region; Build Complexion filter device again, the scene image containing hand region carried out to the segmentation of hand region, obtain hand region and refer to: comprise the following steps:

(1.1) capturing scenes information, and calculate dense optical flow information corresponding with it;

(1.2) travel through Optic flow information, find maximum light stream value, and utilize this maximum light stream value, in X-axis and Y direction, Regularization is carried out to all light streams;

(1.3) according to the light stream after regularization, calculate tone and the saturation degree in the region changed along light stream direction, and mark different color values;

(1.4) set threshold value, according to threshold value, the light stream region of variation of color mark is converted into bianry image, in conjunction with logical operation and mathematical morphological operation, obtain the scene image containing hand region.

(1.5) according to human body complexion Clustering features, structure YCbCr Complexion filter device, eliminate redundancy color and monochrome information;

(1.6) after YCbCr Complexion filter device being filtered, sort descending is carried out to all profiles of the scene image containing hand region, find largest connected region as hand region, realize the segmentation scene image containing hand region being carried out to hand region.

3. Fingertip Detection under complex environment according to claim 1, is characterized in that: in second step, and described employing homalographic block builds the model of each attitude hand region, determines the area of hand region, to calculate the barycenter of hand region; Again hand region contour is sampled, calculate the distance of all configuration samplings point to barycenter, and calculate average barycenter distance; Then determine extended centroid distance according to the last round of finger tip number detected, and be the center of circle with barycenter, extended centroid is apart from carrying out picture circle for radius; Finally reject the point in circle and the circle wrist area that above contiguous pixels number is maximum, and in the outer point finding the maximum barycenter distance in local of circle, then this point is labeled as finger tip; This is taken turns and detects that the number of finger tip compares with the last round of finger tip number detected, if both are equal, then realize the detection of finger tip, otherwise this is taken turns and detects that the number of finger tip is updated to the finger tip number detected, determine extended centroid distance and carry out picture circle, the detection continuing finger tip refers to: comprise the following steps:

(2.1) build hand region display model with homalographic block, according to the different attitudes of hand region, obtain the quantity of homalographic block, determine the area of hand region;

(2.2) barycenter of hand region is calculated according to the area of hand region and the coordinate of each homalographic block;

(2.3) intensive sampling is carried out to hand region contour, calculate the distance of point to barycenter of all samplings, and calculate average barycenter distance;

(2.4) according to the finger tip number N determination extended centroid distance D detected _ext, and be the center of circle with barycenter, extended centroid is apart from D _extfor radius carries out picture circle; And reject the point be included in this circle; Wherein, the initial value of N is set to 0;

(2.5) add up the number of pixels of this circle through hand region, and determine that the region that the upper contiguous pixels number of circle is maximum is wrist area, and reject the point that this wrist area comprises;

(2.6) the outer remaining region of circle is found to the point of the maximum barycenter distance in local, and this point is labeled as finger tip;

(2.7) number N of finger tip will be detected ₁compare with the finger tip number N detected in step (2.4), if both are equal, then realize the detection of finger tip, obtain number and the position of finger tip; Otherwise will the number N of finger tip be detected ₁be updated to the finger tip number N detected in step (2.4), and return step (2.4), continue the detection of finger tip.

4. Fingertip Detection under complex environment according to claim 3, it is characterized in that: in step (2.2), the barycenter that the described area according to hand region calculates hand region with the coordinate of each homalographic block refers to: the barycenter (x being calculated hand region by formula (1) _g, y _g):

$x_g=\frac{s\underset{i=1}{\overset{n}{\&Sigma;}}{x}_i}{s\underset{i=1}{\overset{n}{\&Sigma;}}1}=\frac{\underset{i=1}{\overset{n}{\&Sigma;}}{x}_i}{n},y_g=\frac{s\underset{i=1}{\overset{n}{\&Sigma;}}{y}_i}{s\underset{i=1}{\overset{n}{\&Sigma;}}1}=\frac{\underset{i=1}{\overset{n}{\&Sigma;}}{y}_i}{n}---\left(1\right)$

Wherein, S is the area of hand region, x _i, y _ithat i-th homalographic block is at X-direction and Y direction coordinate respectively.

5. Fingertip Detection under complex environment according to claim 3, it is characterized in that: in step (2.3), the point of all samplings of described calculating to the distance of barycenter, and calculates average barycenter apart from referring to: the point (x calculating all samplings _{point_i}, y _{point_i}) to the Euclidean distance of barycenter, average barycenter distance formula (2) calculates:

$D_{avg}=\frac{\&Integral;\&Integral;\sqrt{{(x_{po\mathrm{int}\_i}-x_g)}^2+{(y_{po\mathrm{int}\_i}-y_g)}^2}dxdy}{\&Integral;\&Integral;dxdy}---\left(2\right)$

Wherein, (x _g, y _g) be the barycenter of hand region.

6. Fingertip Detection under complex environment according to claim 3, is characterized in that: in step (2.4), and the finger tip number N determination extended centroid that described basis has detected is apart from D _extrefer to: the described finger tip number N that detected and extended centroid are apart from D _extthe condition met is:

0≤N<3,D _ext＝1.5×D _avg

3≤N≤5,D _ext＝1.2×D _avg

If the finger tip number N >5 detected, then return the first step; Wherein, D _avgfor average barycenter distance.

7. Fingertip Detection under complex environment according to claim 3, it is characterized in that: in step (2.6), in step (2.6), the described point outer remaining region of circle being found to the maximum barycenter distance in local, and this point is labeled as finger tip refers to: by traversal finger tip point, when the barycenter of 10 point continuous in point is apart from when increasing gradually, record current maximum, when barycenter apart from continuous 10 point barycenter apart from start to reduce time, using the current maximal value recorded as the maximum barycenter distance in local, the maximum barycenter in local is then labeled as finger tip apart from corresponding point.