CN101540001A - Soccer robot projection recognition algorithm - Google Patents
Soccer robot projection recognition algorithm Download PDFInfo
- Publication number
- CN101540001A CN101540001A CN 200910137151 CN200910137151A CN101540001A CN 101540001 A CN101540001 A CN 101540001A CN 200910137151 CN200910137151 CN 200910137151 CN 200910137151 A CN200910137151 A CN 200910137151A CN 101540001 A CN101540001 A CN 101540001A
- Authority
- CN
- China
- Prior art keywords
- search
- projection
- robot
- point
- value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Manipulator (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a soccer robot projection recognition algorithm; compared with other methods, the algorithm has the advantages of less calculation amount, quick image processing speed, easy realization, and the like. The concrete realization method comprises the steps of firstly dividing the site of a competition image into a plurality of areas and searching a seed point in the video image areas, later implementing search in a target area along vertical or horizontal direction by taking the point as a starting point, and then calculating the projection value of the point, recognizing a vertex of a target (or the spherical center of a small ball) according to the projection value, and further recognizing the target; and finally extracting the position and direction of a moving target. The algorithm can be directly applied to an actual game, can quickly and accurately recognize the target, and can meet the needs of soccer robot competition systems with different functions on different application occasions.
Description
Technical field
The invention belongs to image processing field, be specifically related to a kind of Soccer robot projection recognition algorithm.
Background technology
Robot soccer competition is the world cup race by organization arrangement of FIRA's (being called for short FIRA), since 96 years annual one, held in a plurality of famous cities in the world that comprises the BeiJing, China, become one of research focus of hi-tech industry gradually, be subjected to the attention of countries in the world.The Soccer robot system mainly is divided into hardware, sensing (vision), communication, control and master system, and the focus of research mainly is Soccer robot vision research, decision rule research, hardware platform research, communication research etc.Video acquisition module is used for being implemented under the Windows environment, obtain video information by capture card and CCD camera, to be used for carrying out Video processing, its groundwork is quick identification 6 robots and 1 bead between ourselves and the enemy, image information by the capture card acquisition, directly leave in the internal memory, by file, other program can directly not carried out data processing.The size of image is decided according to the performance of capture card.Size of images is a benchmark with the maximum images acquired size of CCD, three class can be set: full-size, maximum sized 1/2, maximum sized 1/4.Video processing and identification module mainly comprise essential information setting, video environment controlling factors, video triggering control on opportunity, color treatments and video tracking identification etc.Color is a kind of sensation of people's vision system to the light signal generating that received, and wavelength optical signals brings the people different color impressions.In the Soccer robot visual identifying system, because color image information is more outstanding, make full use of characteristic more easy to identify and processing of colouring information to video data, therefore in identifying, mainly utilize their colouring information.The design of robot colour code is varied, and team member's pattern also is not quite similar, but color category has certain regulation, and it is as shown in the table in a kind of possible color selection.
Table 1 colouring information table
Aspect Target Recognition and tracking, present recognition methods recognition speed is the principal element in restriction Soccer robot information handling system reaction time always, Chinese patent CN200510027280.6 discloses a kind of visual quick identifying method for football robot, this method is in the robot target identifying, be target at first with ball and team of robot colour code, more than a two field picture a bit search element in the rectangular coil mode for starting point, after searching the robot badge, be center search robot number and direction colour code again with the badge; The defective of this method is that the operation time of spiral search sometimes is longer, and mistake is discerned or trace information is lost if occur, and will strengthen search time at double.
Summary of the invention
Defective at existing Soccer robot target identification method, the present invention proposes a kind of fast video recognizer, this algorithm utilizes the region projection method to catch the destination image data that per second is not less than 30 frames, the big I of robot is provided with input by system, as can be 75 x, 75 x 75mm, be square in image, the foursquare length of side in such cases is 75mm.The specific implementation method is: at first competition area have been divided into several regions, internal point of search in these zones, putting with this then is that starting point is searched in the target area along horizontal or vertical direction, calculate the projection value of this point, identify a summit (or centre of sphere of bead) of target according to these projection values, and then identify target; At last, extract the position and the direction of moving target; Described algorithm can directly apply to real racetrack, and recognition objective rapidly and accurately can satisfy the robot soccer competition system of different application occasion, difference in functionality.Algorithm steps of the present invention is:
The first step is sought seed points
Because the distinctive mark of moving target is a color, in the color chained list of storage, choose color of object value (ball or robot), according to threshold range, in image, search this type of color region, find out seed points.
During the games, choosing except judging next step the zone that may occur according to previous position of seed points also will be according to the characteristics of inter frame image similarity, by the particular location of previous frame even last several two field pictures, the position that target of prediction may occur.After prediction is finished, according to the prediction conclusion, in the estimation range, with the regional center is the center, with robot square inscribe diameter of a circle be the length of side as the region of search, and in this zone according to this color threshold range searching seed points, if search for successfully, return seed points.
If failure outside this region of search, is formed 8 8-neighborhood zones with onesize region of search, and searched for the zone that does not have search successively according to threshold range, the above-mentioned algorithm of recurrence can be obtained seed points.
In second step, the target location is determined
Found seed points, utilize the sciagraphy just can be in the hope of the coordinate figure on a foursquare summit, so, utilize its length of side information and the colouring information (color of robot should be the non-ball look of non-background colour) can be, thereby identify positional information and team member's feature of robot in the hope of the coordinate figure and the robot colour code on other summits.
The 3rd step, target velocity and direction of motion
Utilize the angle at team member's color center and square center can draw the deflection of our robot,, can calculate the information such as speed, direction and future trend of robot according to the judged result of former two field pictures.
Target Recognition Algorithms involved in the present invention can directly apply to real racetrack, recognition objective rapidly and accurately.
Description of drawings
Further specify the present invention below in conjunction with the drawings and specific embodiments.
Fig. 1 is a first two field picture video tracking process flow diagram of the present invention.
Fig. 2 asks the left margin intersection point search routine figure of robot for of the present invention by seed points.
Fig. 3 is a robot of the present invention position fixing process synoptic diagram.
Embodiment
Consult shown in Figure 1ly, the first frame robot graphics video tracking submodule of projection recognition algorithm of the present invention is at first sought non-background colour or non-Chinese red by scan line, obtains seed points P after finding; Ask the foursquare summit of robot by a P with sciagraphy; Robot foursquare other four summits and centers thereof are asked with sciagraphy in the summit thus; Information such as the classification of calculating and differentiation robot and direction; Final we three team members' of robot of output position and directional information.
Consult shown in Figure 2ly, this figure has described the search left margin intersection point submodule of asking in the robot square vertices process of the present invention, and target is to seek the left margin intersection point of current scan line and robot.According to step-size in search b (b＞=1), search for left, judge the color value of current point.For same robot, its color value has carried out normalized, promptly thinks in the contained square area of this robot, and color value is regarded as 1; And not the zone in point, all can regard its color value as 0; If color value is 1, then do projection to Y-axis, continue search; Otherwise, according to criterion output boundary point.
The step of calculating projection is:
1) upwards do y axial projection from current point, projection ultimate range is
The length of side doubly;
2) method of calculating projection value is that if color value is 1, this point is still internal point, adds up 1 from the upwards search of current point, otherwise, if color value is 0, the outside has been described, stop search the preservation accumulated value.The search end condition is the projector distance of upwards searching for.Similar end condition has been arranged, just can handle the situation that several robots stack well.
3) search for and calculate projection left, in search procedure left, projection value successively decreases or increases progressively, and this read-me is just at the square interior normal searching, if current some color value is 0, step-size in search is return back.Judge this current point of back, if color value is non-vanishing, the change step-size in search is 1, continues search left, at this moment, have only apart from frontier point L one step away, search 0 value after, promptly found border intersection point L, it is exactly the previous point on the searching route.
Consult shown in Figure 3ly, this figure has described the searching algorithm of asking the robot square vertices of the present invention.By specifying upwards (or downwards) search of step-length b, the maximum length of search is the length of side from frontier point L
Doubly (if the overlapping situation of similar robot is arranged, it is too far away to be unlikely to enter non-recognizate inside) is 0 until color value, and another border has been arrived in expression.(or may several steps, this moment the main robot adhesion situation of distinguishing) color value is 0 if one step of the beginning of upwards search, the out-of-bounds that express possibility, and the summit that search for should be searched for downwards necessarily not below, upwards search of hypothesis here.In search procedure from top to bottom, whenever move up a step, following two steps are arranged:
1) checks current some color value, judge whether to finish.
2) from current point as projector distance be left
The transverse projection of the length of side, projection value adds up.
Judge at last and confirm projection value from little change big after, the separatrix from diminishing greatly again, the most left color value is that 1 point is summit S on this separatrix.
Target projection recognizer of the present invention has that amount of calculation is few, image processing speed is fast and be easy to reality Now wait advantage, can directly apply to real racetrack, can identify rapidly and accurately target, can satisfy different application The robot soccer competition system of occasion, difference in functionality.
Claims (8)
1, a kind of Soccer robot projection recognition algorithm, the steps include: at first the place of tournament picture to be divided into several regions, by pressing the line scanning video image, seed points of search in these zones, putting with this then is that starting point is searched in the target area along horizontal or vertical direction, calculate the projection value of this point, identify a summit (or centre of sphere of bead) of target, and then identify the foursquare all summits of Soccer robot according to projection value; At last, extract the position and the direction of moving target.
2, according to claim 1 described Soccer robot projection recognition algorithm, it is characterized in that: the choosing of described seed points except judging next step the zone that may occur according to previous position, also will be according to the characteristics of inter frame image similarity, by the particular location of previous frame even last several two field pictures, the position that target of prediction may occur; After prediction is finished,,, be the center with the regional center in the estimation range according to the prediction conclusion, with robot square inscribe diameter of a circle be the length of side as the region of search, and in this zone according to this color threshold range searching seed points.
3, according to claim 1 described Soccer robot projection recognition algorithm, it is characterized in that: the calculation procedure of certain spot projection is in the described image:
1) upwards do y axial projection from current point, projection ultimate range is
The length of side doubly;
2) for same robot, its color value carries out normalized, promptly thinks in the contained square area of this robot, and color value is regarded as 1; And not the zone in point, all can regard its color value as 0;
3) method of calculating projection value is from the upwards search of current point, if color value be 1 and this point be still internal point, add up 1, otherwise, if color value is 0, the outside has been described, stop search, preserve accumulated value.
4, according to claim 1 described Soccer robot projection recognition algorithm, it is characterized in that: the method on a described summit according to the projection value recognition objective be for: at first from seed points, search left, seek the left margin intersection point of scan line and robot, then according to this left margin intersection point, final search also obtains a summit of robot.
5, according to claim 1 described Soccer robot projection recognition algorithm, it is characterized in that: four summits of described Soccer robot and direction can be obtained by calculating by the length of side and the colouring information of a summit that has obtained, Soccer robot.
6, according to claim 1 described Soccer robot projection recognition algorithm, it is characterized in that: the speed of described Soccer robot and future trend can obtain according to the positional information calculation of former two field pictures.
7, according to claim 4 described Soccer robot projection recognition algorithms, it is characterized in that: during the left margin intersection point of described searching scan line and robot, search for and calculate projection left, if in search procedure left, projection value successively decreases or increases progressively, this read-me is just at the square interior normal searching, if current some color value is 0, step-size in search is return back; Judge this current point of back, if color value is non-vanishing, the change step-size in search is 1, continues search left, search 0 value after, promptly found border intersection point L, it is exactly the previous point on the searching route.
8, according to claim 4 described Soccer robot projection recognition algorithms, it is characterized in that: in the process on a summit of described searching machine people, by specifying upwards (or downwards) search of step-length, the maximum length of search is the length of side from frontier point
Doubly, be 0 until color value, another border has been arrived in expression; (or may several steps, this moment the main robot adhesion situation of distinguishing) color value is 0 if one step of the beginning of upwards search, the out-of-bounds that express possibility, and the summit that search for should be searched for downwards necessarily not below; Here upwards search of hypothesis.In search procedure from top to bottom, whenever move up a step, following two steps are arranged:
1) checks current some color value, judge whether to finish;
2) from current point as projector distance be left
The transverse projection of the length of side, projection value adds up;
Judge at last and confirm projection value from little change big after, the separatrix from diminishing greatly again, the most left color value is that 1 point is summit S on this separatrix.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200910137151 CN101540001A (en) | 2009-05-04 | 2009-05-04 | Soccer robot projection recognition algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200910137151 CN101540001A (en) | 2009-05-04 | 2009-05-04 | Soccer robot projection recognition algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101540001A true CN101540001A (en) | 2009-09-23 |
Family
ID=41123182
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 200910137151 Pending CN101540001A (en) | 2009-05-04 | 2009-05-04 | Soccer robot projection recognition algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101540001A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102542294A (en) * | 2011-12-29 | 2012-07-04 | 河海大学常州校区 | Centralized control type soccer robot identification system and identification method for double visual information fusion |
| CN103301630A (en) * | 2013-06-16 | 2013-09-18 | 西安科技大学 | Coordination and cooperation control method for football robots and system thereof |
| CN110618682A (en) * | 2019-09-24 | 2019-12-27 | 河海大学常州校区 | Color code structural design and recognition algorithm of centralized control type football robot |
| CN114770555A (en) * | 2022-05-24 | 2022-07-22 | 乐聚(深圳)机器人技术有限公司 | Robot and robot control method |
-
2009
- 2009-05-04 CN CN 200910137151 patent/CN101540001A/en active Pending
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102542294A (en) * | 2011-12-29 | 2012-07-04 | 河海大学常州校区 | Centralized control type soccer robot identification system and identification method for double visual information fusion |
| CN103301630A (en) * | 2013-06-16 | 2013-09-18 | 西安科技大学 | Coordination and cooperation control method for football robots and system thereof |
| CN103301630B (en) * | 2013-06-16 | 2015-06-10 | 西安科技大学 | Coordination and cooperation control method for football robots and system thereof |
| CN110618682A (en) * | 2019-09-24 | 2019-12-27 | 河海大学常州校区 | Color code structural design and recognition algorithm of centralized control type football robot |
| CN110618682B (en) * | 2019-09-24 | 2022-11-01 | 河海大学常州校区 | Centralized control type football robot color code structure and identification method thereof |
| CN114770555A (en) * | 2022-05-24 | 2022-07-22 | 乐聚(深圳)机器人技术有限公司 | Robot and robot control method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Yang et al. | Real-time face detection based on YOLO | |
| CN108873943B (en) | Image processing method for centimeter-level accurate landing of unmanned aerial vehicle | |
| WO2019179441A1 (en) | Focus tracking method and device of smart apparatus, smart apparatus, and storage medium | |
| CN107301665B (en) | Depth camera with variable-focus optical camera and control method thereof | |
| Breitenmoser et al. | A monocular vision-based system for 6D relative robot localization | |
| US11468682B2 (en) | Target object identification | |
| CN105046206B (en) | Based on the pedestrian detection method and device for moving prior information in video | |
| CN106054627B (en) | Control method and device based on gesture recognition and air conditioner | |
| CN110427797B (en) | Three-dimensional vehicle detection method based on geometric condition limitation | |
| CN109341692B (en) | Line navigation method and robot | |
| CN112825192B (en) | Object identification system and method based on machine learning | |
| CN109725645A (en) | Nested unmanned aerial vehicle landing cooperation sign design and relative pose acquisition method | |
| CN114415736B (en) | Multi-stage visual accurate landing method and device for unmanned aerial vehicle | |
| CN109584264B (en) | Unmanned aerial vehicle vision guiding aerial refueling method based on deep learning | |
| CN101540001A (en) | Soccer robot projection recognition algorithm | |
| CN113377118A (en) | Multi-stage accurate landing method for unmanned aerial vehicle hangar based on vision | |
| Zelener et al. | Cnn-based object segmentation in urban lidar with missing points | |
| CN110163916B (en) | Rolling target-based surrounding type circular mark point matching method | |
| CN114612786A (en) | Obstacle detection method, mobile robot and machine-readable storage medium | |
| JP2023043632A (en) | Information processor, program, and method for positioning | |
| Rao et al. | Real time vision-based autonomous precision landing system for UAV airborne processor | |
| Wang et al. | Research on target detection and recognition algorithm based on deep learning | |
| Makrigiorgis et al. | How High can you Detect? Improved accuracy and efficiency at varying altitudes for Aerial Vehicle Detection | |
| Duan et al. | Image digital zoom based single target apriltag recognition algorithm in large scale changes on the distance | |
| CN114463664B (en) | Novel ice ball tracking method for ice ball movement |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Open date: 20090923 |