CN101751118B - Object endpoint positioning method and application system - Google Patents

Abstract

A method for locating the endpoints of an object is used to locate the end positions of two limbs of an object. In this method, the original image obtained is subjected to foreground processing to obtain a foreground image. The foreground image corresponds to the object contour in the original image. A plurality of turning points are obtained based on the foreground image, and the connection of the turning points forms a polygonal curve close to the object contour. The turning points are classified into convex points or concave points according to the angles thereof, and a plurality of selected convex points and a plurality of selected concave points are selected. Two of the selected convex points are selected as two temporary endpoints. The connection of the two selected temporary endpoints and a selected concave point located between the two temporary endpoints forms a triangle corresponding to the two limbs of the object in the original image. Two positioning endpoints are determined based on the two temporary endpoints to locate the end positions of the two limbs of the object.

Description

Object endpoint positioning method and application system

technical field

The present invention relates to an object endpoint positioning method and an applied system, and in particular relates to an object endpoint positioning method and an applied system for locating the end positions of two limbs of an object.

Background technique

A human-computer interaction interface is an interface that allows a human to interact with a computer. Generally speaking, the human-computer interaction interface usually includes a keyboard, a mouse, or a touch panel. In such an interface, a person is called a user. Through the human-computer interaction interface, the user can achieve the purpose of controlling the computer or interacting with the computer.

In the patent disclosed in U.S. Patent No. 5,524,637, an interactive system for measuring physiological exercise ("interactive system for measuring physiological exercise") is proposed, which is to wear an acceleration sensor (accelerometer) on the user's feet , and make the user stand on a pressure sensor board (pressure sensor) to judge the force exerted by the user's feet and the moving acceleration of the feet. In this way, the user can interact with the interactive system through the force exerted and the moving speed of the feet.

Furthermore, in recent years, some researchers have used image processing technology to increase the interaction between users and computers. In the patent disclosed in U.S. Patent No. 6,308,565, a system and method for tracking and assessing movement skills in multidimensional space ("system and method for tracking and assessing movement skills in multidimensional space") is proposed, which uses active or passive identification The (marker) is pasted on the user's feet, and image processing is used to detect the movement of the mark to determine the coordinate position and movement state of the user's feet in space. In this way, the user can interact with the interactive system by moving the feet.

Although a variety of human-computer interaction interfaces have been proposed, in the above-mentioned embodiments, if the user-computer interaction interface is to be able to determine the end positions of the user's feet, the user must wear a special device or Clothing (such as the above-mentioned accelerometer, and signs). In this way, it will cause inconvenience to the user, and may also reduce the willingness of the user to use. Therefore, how to locate the end positions of the user's feet without causing inconvenience to the user is still one of the issues that the industry is working on.

Contents of the invention

The object of the present invention is to provide a method for locating an object endpoint and an applied system, which can locate the terminal positions of two limbs of an object. The two limbs of this object can be the double feet or two fingers of human body. With the method for locating the endpoint of the object and its application, the user does not need to wear special devices or clothing. In this way, the user's convenience of use can be improved.

To achieve the above object, according to the first aspect of the present invention, a method for locating the endpoints of an object is proposed, which is used for locating the endpoints of two limbs of an object. This method includes the following steps. First, an original image is obtained, and the original image has image information corresponding to an object. Next, foreground processing is performed on the original image to obtain a foreground image. The foreground image corresponds to the outline of the object. Then, a plurality of turning points are obtained according to the foreground image, and the connection of turning points can form a polygonal curve which is substantially similar to the contour of the object in the original image. Afterwards, according to each turning point and the angle formed by the corresponding adjacent two turning points, a plurality of convex points and a concave point are determined from these turning points, and a plurality of selected convex points and concave points are selected along a predetermined direction. Multiple selected pits. Again, select two of these selected convex points as two tentative endpoints, the connection between the selected two tentative endpoints and a selected concave point between the two tentative endpoints forms the connection with the object in the original image A triangle corresponding to the two limbs. Finally, two positioning endpoints are determined according to the two provisional endpoints to locate the end positions of the two limbs of the object.

According to a second aspect of the present invention, an object endpoint positioning system is provided for locating the endpoints of two limbs of an object. The system includes a capture unit, a processing unit, a matching unit, and a positioning unit. The capturing unit is used for obtaining an original image, and the original image has image information corresponding to the object. The processing unit is used to perform foreground processing on the original image to obtain a foreground image, and the foreground image corresponds to the outline of the object. The processing unit is also used to obtain a plurality of turning points according to the foreground image, and the connection of these turning points can form a polygonal curve which is substantially similar to the outline of the object in the original image. The processing unit is used to determine a plurality of convex points and a plurality of concave points from these turning points according to the angle formed by each turning point and two corresponding adjacent turning points, and select a plurality of selected convex points along a predetermined direction. point and multiple selected concave points. The matching unit is used to select two of the selected convex points as two tentative endpoints, and the connection between the selected two tentative endpoints and a selected concave point between the two tentative endpoints is formed with the original image. The two limbs of the object correspond to a triangle. The positioning unit is used for determining two positioning endpoints according to the two tentative endpoints, so as to locate the terminal positions of the two limbs of the object.

Description of drawings

FIG. 1 is a flowchart of a method for locating an object endpoint according to an embodiment of the present invention.

FIG. 2 is a block diagram of an object endpoint positioning system applying the object endpoint positioning method shown in FIG. 1 .

FIGS. 3-7 respectively illustrate an example of various images generated by the object endpoint positioning system when executing the object endpoint positioning method.

Explanation of main component symbols in the drawings

200 object endpoint positioning system; 210 capture unit; 220 processing unit; 230 matching unit; 240 positioning unit; Direction; F2 outline; F3 polygonal curve; Im1 original image; Im2 foreground image; Px, Py positioning endpoint; S110-S160 process steps; t1, t2 tentative endpoint;

Detailed ways

In order to make the above content of the present invention more comprehensible, preferred embodiments are specifically cited below and described in detail with accompanying drawings.

Please refer to FIG. 1 , which is a flowchart of a method for locating an object endpoint according to an embodiment of the present invention. This method is used to locate the end positions of the two limbs of an object. This method includes the following steps.

First, as shown in step S110, an original image is obtained, which has image information corresponding to the object. Next, as shown in step S120, foreground processing is performed on the original image to obtain a foreground image. The foreground image corresponds to the outline of the object.

Then, in step S130, a plurality of turning points are obtained according to the foreground image. The connection of the turning points can form a polygonal curve substantially similar to the outline of the object in the original image. After that, in step S140, according to the included angle formed by each turning point and the corresponding two adjacent turning points, a plurality of convex points and a plurality of concave points are determined from these turning points, and a plurality of concave points are selected along a predetermined direction. A selected convex point and a plurality of selected concave points.

Next, in step S150, select two of the selected bumps as two tentative endpoints. A connection of the selected two tentative endpoints and a selected concave point between the two tentative endpoints forms a triangle corresponding to the two limbs of the object in the original image. Finally, in step S160, two positioning endpoints are determined according to the two tentative endpoints to locate the terminal positions of the two limbs of the object.

Taking an object endpoint positioning system applying the object endpoint positioning method in FIG. 1 as an example, the detailed description is as follows. Please refer to Figure 2 and Figures 3 to 7 at the same time. FIG. 2 is a block diagram of an object endpoint location system 200 applying the object endpoint location method shown in FIG. 1 . 3-7 illustrate an example of various images generated by the object endpoint positioning system 200 when performing the object endpoint positioning method.

The object endpoint positioning system 200 can locate the terminal position Ft of the human body's feet F, as shown in FIG. 3 . The object endpoint positioning system 200 includes a capture unit 210 , a processing unit 220 , a matching unit 230 , a positioning unit 240 , and a tracking unit 250 .

The capturing unit 210 is used for obtaining an original image Im1. As shown in FIG. 3 , the original image Im1 has image information corresponding to the feet F of the human body.

The processing unit 220 is configured to perform foreground processing on the original image Im1 to obtain a foreground image. When the processing unit 220 performs foreground processing on the original image Im1, for example, edge detection is performed on the original image Im1. In this way, the processing unit 220 will obtain the foreground image Im2 having the edge information of the original image Im1, and the foreground image Im2 will have the outline F2 of both feet, as shown in FIG. 4 .

When the foreground processing is performed on the original image Im1, the obtained foreground image Im2 usually includes the contours of scene objects, such as the contour F2 of the feet and the contours A and B of other objects. Therefore, the processing unit 220 can filter the foreground image Im2 to preserve the outline F2 of the feet. In practice, since the contour F2 of the feet is usually the block with the largest area, the processing unit 220 can find out the contour F2 of the feet and keep it by counting the block areas of the contours F2, A and B. .

Next, as shown in FIG. 5 , the processing unit 220 is further configured to obtain a plurality of turning points c1˜cn according to the foreground image Im2. The connection of these turning points c1˜cn can form a polygonal curve F3, and the shape of the polygonal curve F3 is substantially similar to the outline F2 of both feet.

Afterwards, the processing unit 220 is used to determine a plurality of convex points and a plurality of concave points from the turning points c1-cn according to the included angle formed by each turning point and two corresponding adjacent turning points.

For example, bumps and pits can be defined as follows. The included angle corresponding to the turning point defined as a convex point is an included angle between 0 and 120 degrees; and the included angle corresponding to a turning point defined as a concave point is an included angle greater than 240 degrees. In the above definition, the included angle is the inner angle of the polygonal curve F3. In this way, as shown in Figure 5, the angle formed by the turning point c2 and the corresponding two adjacent turning points c1 and c3 conforms to the definition of a bump, and the angle formed by the turning point c3 and the corresponding two adjacent turning points c2 and c4 The included angle conforms to the definition of a concave point.

For example, after determining a plurality of bumps and a plurality of pits, the processing unit 220 selects four selected bumps a1 - a4 and two selected pits b1 and b2 along a predetermined direction. As shown in FIG. 6 , the predetermined direction is, for example, a direction D1 of the ends of the feet F relative to the top ends of the feet F. As shown in FIG.

Next, the matching unit 230 selects two of the selected bumps a1-a4 as the two tentative endpoints t1 and t2, and the selected two tentative endpoints t1 and t2 are connected to one of the selected bumps between the two tentative endpoints t1 and t2. The connection of the selected pit points forms a triangle corresponding to the human feet F in the original image Im1. As shown in FIG. 7 , two selected convex points a1 and a2 form a triangle with a selected concave point b1 located therebetween, and this triangle essentially locates the positions of the feet F of the human body.

Further, when the matching unit 230 selects two of the selected convex points a1-a4 as the two tentative endpoints t1 and t2, the matching unit 230 judges the selected convex points a1-a4 and the selected concave points b1 And whether b2 conforms to the triangular feature matching.

The so-called triangular feature matching refers to that the matching unit 230 judges whether the online slope of the second of the selected bumps a1 - a4 is less than a predetermined slope according to a vector perpendicular to the predetermined direction D1 . The matching unit 230 also determines whether the position of one of the selected concave points b1 and b2 projected onto the vector is located between the positions of the other two of the selected convex points a1-a4 projected onto the vector.

For example, in FIG. 7 , the aforementioned predetermined slope is, for example, a slope of 45°. The online slope S1 of the selected convex points a1 and a2 is smaller than the predetermined slope, and the selected concave point b1 is projected to the position d1 on the vector D2, which is located at the positions d2 and d3 of the two selected convex points a1 and a2 projected on the vector D2 between. Therefore, the matching unit 230 determines that the selected convex points a1 and a2 and the selected concave point b1 meet the triangular feature matching.

Furthermore, as shown in FIG. 6 , if the side close to the end position Ft of the feet F is defined as downward, the predetermined direction D1 is from downward to upward. After the matching unit 230 determines that the selected convex points a1 and a2 match the selected concave point b1 according to the triangular feature, the matching unit 230 further determines whether the selected convex points a1 and a2 are lower than the selected concave point b1. In the example shown in FIG. 7 , the matching unit 230 will determine that the two selected convex points a1 and a2 are lower than the selected concave point b1 .

Further, when the matching unit 230 judges whether the selected convex points a1-a4 and the selected concave points b1 and b2 meet the triangular feature matching process, if the matching unit 230 determines that the selected convex points a1 and a2 and the selected concave points In addition to b1 conforming to the triangular feature matching, it is also determined that the other two selected convex points and the corresponding selected concave point also conform to the triangular feature matching. At this time, the matching unit 230 is also used to determine whether the area formed by the selected convex points a1 and a2 and the selected concave point b1 is larger than the area formed by the other two selected convex points and a corresponding selected concave point.

In the example shown in FIG. 7 , assuming that the matching unit 230 determines that the selected convex points a1 and a2 and the selected concave point b1 conform to the triangular feature matching, and determines whether the selected convex points a1 and a2 are closer to the lower side than the selected concave point b1 , It is also determined that the area formed by the selected convex points a1 and a2 and the selected concave point b1 is the largest. In this way, the matching unit 230 decides to select the two selected bumps a1 and a2 as the two tentative endpoints t1 and t2 .

Please refer to the aforementioned FIG. 2, after the matching unit 230 uses the selected bumps a1 and a2 as the two tentative endpoints t1 and t2, then the positioning unit 240 is used to determine the two positioning endpoints Px and Px according to the two tentative endpoints t1 and t2. Py. It can be seen from FIG. 7 that the positions of the selected bumps a1 and a2 determined by the matching unit 230 are substantially the end positions Ft of the feet F. As shown in FIG. In this way, the two positioning endpoints Px and Py determined by the positioning unit 240 can locate the terminal positions Ft of the feet F of the human body.

In the above description, the object end point positioning system 200 analyzes and judges the information of the image in an image processing manner to locate the end positions of the feet of the human body. Compared to systems that require the user to wear special devices or clothing to position the human body's feet. The object endpoint positioning system 200 of the present invention does not require the user to wear special devices or clothing. In this way, the convenience of use of the user can be improved, and there is no problem of reducing the willingness of the user to use.

In addition, when the matching unit 230 determines that the selected convex points a1 - a4 and the selected concave points b1 and b2 do not meet the triangular feature matching, the matching unit 230 will enable the tracking unit 250 . The enabled tracking unit 250 obtains two tracking endpoints t1' and t2' according to the two previous endpoints. The two previous end points are the end positions of the two limbs of the object located by the positioning unit 240 in a previous original image.

More specifically, the tracking unit 250 can track the end positions of both feet located in the previous original image to generate two tracking end points t1' and t2' which are close to the actual two positioning end points. In this way, even if the positioning unit 240 cannot correctly determine the two positioning endpoints Px and Py through the matching unit 230, the positioning unit 240 can also determine the actual two tracking endpoints t1' and t2' according to the tracking unit 250. Two positioning endpoints Px and Py that are close to the positioning endpoint. In this way, the operation stability of the object endpoint positioning system 200 can be improved.

In the multiple original images sequentially acquired by the capture unit 210, since the moving distance of the object (for example: the feet F of the human body) will be within a certain range, the position of the end positions of the two limbs located by the positioning unit 240 The change (for example: the change of the end position Ft of the feet F) will also be within a certain range. Therefore, if the matching unit 240 cannot locate the end position Ft of the feet F of the human body in a certain original image, the tracking unit 250 can track the end position Ft of the feet F located in the previous original image to Find two tracking endpoints t1' and t2' that are close to the actual two positioning endpoints. How the tracking unit 250 obtains the two tracking endpoints t1' and t2' is described with several examples.

In the first example, when the tracking unit 250 obtains the two tracking endpoints t1' and t2', the tracking unit 250 determines the two endpoints according to the brightness changes of the surrounding pixels of the two previous endpoints in the previous original image and the original image Im. Trace endpoints t1' and t2'.

In the second example, when the tracking unit 250 obtains the two tracking endpoints t1' and t2', the tracking unit 250 determines the two endpoints according to the color changes of the surrounding pixels of the two previous endpoints in the previous original image and the original image Im. Trace endpoints t1' and t2'.

In the third example, when the tracking unit 250 obtains the two tracking endpoints t1' and t2', the tracking unit 250 decides in a predictive or probabilistic manner according to the positions represented by the two previous endpoints and the other two previous endpoints respectively. Two tracking endpoints t1' and t2'. The other two previous endpoints are the end positions of the two limbs of the object positioned by the positioning unit 240 in another previous original image, and the original image Im, the previous original image and another previous original image are successively acquired by the capture unit 210 .

In the above-mentioned first and second examples, the two tracking endpoints t1' and t2' are determined respectively by the brightness and color changes in the original image Im and the previous original image. In the third example, the two tracking endpoints t1' and t2' are determined by prediction or probability. These examples are used to illustrate the present invention, not to limit the present invention. As long as the matching unit 230 determines that the selected convex points a1-a4 and the selected concave points b1 and b2 do not meet the triangular feature matching, the tracking unit 250 can be used to generate two tracking endpoints t1' and t2' that are close to the actual two positioning endpoints , are all within the protection scope of the present invention.

The object endpoint positioning system 200 of the present invention can determine two positioning endpoints Px and Py after obtaining an original image Im1 to locate the terminal positions Ft of the human feet F. Furthermore, if the object endpoint positioning system 200 can sequentially obtain multiple original images in a period of time, it can sequentially locate the end positions of the human feet from these original images. In this way, the object endpoint positioning system 200 can further analyze the position information to determine the moving direction, moving speed, or action pattern of the feet F of the human body during this period.

Furthermore, the above description is based on an example where the object endpoint positioning system 200 can locate the terminal positions Ft of the feet F of the human body, but it is not limited thereto. The object end point positioning system 200 of the present invention can also locate the end positions of the two fingers of the human body. At this time, the above-mentioned predetermined direction will be the direction of the tip of the finger relative to the tip of the finger. That is, at this time, the object endpoint positioning system 200 selects the selected convex point and the selected concave point along the direction of the tip of the finger relative to the tip of the finger. In this way, the object endpoint positioning system 200 can also select two selected convex points as two tentative endpoints, and the connection between the two tentative endpoints and a selected concave point in between forms a triangle corresponding to two fingers, thereby positioning out the end position of the two fingers.

The present invention can locate the end positions of two limbs of an object by using the object end point positioning method described in the above embodiments and the applied system. The two limbs of this object can be the double feet or two fingers of human body. With the method for locating the endpoint of the object and its application, the user does not need to wear special devices or clothing. In this way, the convenience of use of the user can be improved, and there is no problem of reducing the willingness of the user to use.

In summary, although the present invention has been described above with preferred embodiments, it is not intended to limit the present invention. Those skilled in the art may make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention should be determined by the content defined by the scope of the claims of the application.

Claims (20)

1. object end positioning method, in order to the terminal position of two limbs of locating an object, this method comprises:

Obtain a raw video, this raw video has the image information corresponding to this object;

This raw video is carried out perspective process, and to obtain a prospect image, this prospect image corresponds to the profile of this object;

According to this prospect image, obtain a plurality of turning points, those turning points online form with this raw video in the close in fact polygon curve of profile of this object;

According to this turning point respectively and corresponding two adjacent formed angles of turning point, determining a plurality of salient points and a plurality of concave point from those turning points, and select a plurality of selected salient points and a plurality of selected concave point along a predetermined direction;

Its that select those selected salient points be two as two tentative end points, the corresponding triangle of these two limbs of this object in this two tentative end points selected and the online formation of a selected concave point between this two tentative end points and this raw video; And

Determine two location end points according to this two tentative end points, locate the terminal position of these two limbs of this object.

The method of claim 1, wherein this predetermined direction be relevant to this object in this raw video the end of these two limbs with respect to the direction on the top of these two limbs.

3. the method for claim 1, wherein select before this its two steps as these two tentative end points of those selected salient points, this method comprises:

Judge whether those selected salient points and those selected concave points meet the triangle characteristic matching; And

If those selected salient points this its two and this one of which of those selected concave points meet the triangle characteristic matching, then determine with those selected salient points this its two as this two tentative end points.

4. method as claimed in claim 3 wherein, comprises in this its two steps as these two tentative end points of decision with those selected salient points:

If those selected salient points this its two and this one of which of those selected concave points meet the triangle characteristic matching; And those selected salient points in addition two and another of those selected concave points also meet the triangle characteristic matching; Then judge those selected salient points this its two and the formed area of this one of which of those selected concave points; Whether greater than those selected salient points this in addition two and this another formed area of those selected concave points; If those selected salient points this its two and the formed area of this one of which of those selected concave points bigger, then carry out this decision with those selected salient points this its two as this two fix tentatively end points steps.

5. method as claimed in claim 3, wherein, judge that the step whether those selected salient points and those selected concave points meet the triangle characteristic matching comprises:

Vector according to vertical this predetermined direction; Whether this its online slope of two of judging those selected salient points is less than a predetermined slope; Whether and this one of which of judging those selected concave points is projected to the position on this vector, between this its two positions of being projected on this vector of those selected salient points.

6. method as claimed in claim 3; Wherein, If the side near the terminal position of this two limbs of this object is defined as the below, then this predetermined direction is the direction by below to top, and decision comprises this its two steps as these two tentative end points of those selected salient points:

Judge those selected salient points this its two whether than this one of which of those selected concave points near the below, if then carry out this its two the steps as this two tentative end points of this decision with those selected salient points.

7. method as claimed in claim 3 comprises:

Obtain two according to two first forward terminals and follow the trail of end points, this two first forward terminal is the terminal position of two limbs of this object of in a previous raw video, being located;

Wherein when those selected salient points and those selected concave points did not meet the triangle characteristic matching, then the step according to this this two location end points of two tentative end points decisions replaced with:

According to this two trackings end points, decide this two location end points.

8. method as claimed in claim 7, wherein, the step that obtains this two trackings end points comprises:

In this previous raw video and this raw video, the brightness that accordings to the surrounding pixel of this two first forward terminal changes, and decides this two trackings end points.

9. method as claimed in claim 7, wherein, the step that obtains this two trackings end points comprises:

In this previous raw video and this raw video, according to the change color of the surrounding pixel of this two first forward terminal, decide this two trackings end points.

10. method as claimed in claim 7, wherein, the step that obtains this two trackings end points comprises:

According to this two first forward terminal and the two first forward terminals institute position of representative respectively in addition,, decide this two trackings end points with the mode of prediction or probability;

Wherein, two first forward terminals lie in the terminal position of two limbs of this object of being located in another previous raw video in addition, and this raw video, this previous raw video and this another previous raw video are obtained continuously.

11. an object end positioning system, in order to the terminal position of two limbs of locating an object, this system comprises:

One acquisition unit, in order to obtain a raw video, this raw video has the image information corresponding to this object;

One processing unit; In order to this raw video is carried out perspective process; To obtain a prospect image, this prospect image corresponds to the profile of this object, and this processing unit is in addition in order to obtain a plurality of turning points according to this prospect image; Those turning points online form with this raw video in the close in fact polygon curve of profile of this object; This processing unit is in order to according to this turning point respectively and corresponding two adjacent formed angles of turning point, determining a plurality of salient points and a plurality of concave point from those turning points, and selects a plurality of selected salient points and a plurality of selected concave point along a predetermined direction;

One matching unit; In order to select those selected salient points its two as two tentative end points, the corresponding triangle of these two limbs of this object in this two tentative end points selected and the online formation of a selected concave point between this two tentative end points and this raw video; And

Locating unit in order to determine two location end points according to this two tentative end points, is located the terminal position of these two limbs of this object.

12. system as claimed in claim 11, wherein, this predetermined direction is to be relevant to the end of these two limbs of this object in this raw video with respect to the direction on the top of these two limbs.

13. system as claimed in claim 11; Wherein, When this matching unit select those selected salient points this its two during as this two tentative end points; This matching unit judges whether those selected salient points and those selected concave points meet the triangle characteristic matching, if this matching unit judge those selected salient points this its two and this one of which of those selected concave points meet the triangle characteristic matching, then this matching unit decision select those selected salient points this its two as this two tentative end points.

14. system as claimed in claim 13; Wherein, If this matching unit judge those selected salient points this its two and this one of which of those selected concave points meet the triangle characteristic matching; And judge those selected salient points in addition two and another of those selected concave points also meet the triangle characteristic matching; Then this matching unit in order to judge those selected salient points this its two and the formed area of this one of which of those selected concave points; Whether greater than those selected salient points this in addition two and this another formed area of those selected concave points, if those selected salient points this its two and the formed area of this one of which of those selected concave points bigger, then this matching unit decision with those selected salient points this its two as this two tentative end points.

15. system as claimed in claim 13; Wherein, When this matching unit judged whether those selected salient points and those selected concave points meet the triangle characteristic matching, according to perpendicular to a vector of a predetermined direction, whether this its online slope of two of judging those selected salient points was less than a predetermined slope; Whether and this one of which of judging those selected concave points is projected to the position on this vector, between this its two positions of being projected on this vector of those selected salient points.

16. system as claimed in claim 13; Wherein, If the side near the terminal position of this two limbs of this object is defined as the below, then this predetermined direction is the direction by below to top, and those selected salient points this its two than this one of which of those selected concave points near the below.

17. system as claimed in claim 13, wherein, this matching unit comprises:

One tracing unit is followed the trail of end points, the terminal position of two limbs of this object that this two first forward terminal is located in a previous raw video for this positioning unit in order to obtain two according to two first forward terminals;

Wherein when this matching unit judged that those selected salient points and those selected concave points do not meet the triangle characteristic matching, this positioning unit replacement determined two location end points according to this two tentative end points, and two followed the trail of end points according to this, decided this two location end points.

18. system as claimed in claim 17; Wherein, when this tracing unit was obtained this two trackings end points, this tracing unit was in this previous raw video and this raw video; The brightness that accordings to the surrounding pixel of this two first forward terminal changes, and decides this two trackings end points.

19. system as claimed in claim 17; Wherein, when this tracing unit was obtained this two trackings end points, this tracing unit was in this previous raw video and this raw video; The change color that accordings to the surrounding pixel of this two first forward terminal decides this two trackings end points.

20. system as claimed in claim 17; Wherein, when this tracing unit is obtained this two when following the trail of end points, this tracing unit is according to this two first forward terminal and the position represented respectively of two first forward terminals institute in addition; With the mode of prediction or probability, decide this two trackings end points;

Wherein, the terminal position of two limbs of this object that these other two first forward terminals are located in another previous raw video for this positioning unit, this raw video, this previous raw video and this another previous raw video are to be obtained by this acquisition unit continuously.

Images