CN102375983A - Image segmentation processing method based on area matching optimization K-means clustering algorithm - Google Patents
Image segmentation processing method based on area matching optimization K-means clustering algorithm Download PDFInfo
- Publication number
- CN102375983A CN102375983A CN2010102547467A CN201010254746A CN102375983A CN 102375983 A CN102375983 A CN 102375983A CN 2010102547467 A CN2010102547467 A CN 2010102547467A CN 201010254746 A CN201010254746 A CN 201010254746A CN 102375983 A CN102375983 A CN 102375983A
- Authority
- CN
- China
- Prior art keywords
- area
- image
- result
- clustering algorithm
- means clustering
- 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.)
- Granted
Links
Images
Abstract
The invention relates to an image segmentation processing method based on an area matching optimization K-means clustering algorithm, comprising steps of firstly extracting vehicle feature points of front and back frame images, and then comparing an area overlapping situation of front and back frame vehicles, extracting positions of the feature points in an area overlapping region and the positions of the rest feature points, respectively calculating a mean value of the two groups of the feature points as two types of initial clustering central points to be segmented, and then implementing K-mean segmentation, correcting a classification situation of the feature points in the area overlapping region according to an output clustering result, and meanwhile, judging whether the clustered vehicles are reasonable or not; and if not, re-clustering the clustering result and recounting clustering centres, ending clustering segmentation until the reasonable vehicles are found, and then feeding back a tracking result. The method is on the basis of area matching optimization, and adopts fixed clustering numbers to implement the segmentation; and vehicle targets obtained by the K-mean segmentation do not need the next round of matching treatment, thereby a processing speed is accelerated, and time is saved.
Description
Technical field
The present invention relates to the video detection technology field, particularly video image target following technical field specifically is meant the image segmentation disposal route based on area matched optimization K-means clustering algorithm.
Background technology
Video frequency vehicle Tracking Recognition technology at first will be partitioned into moving target in video image, then could tracking target.Under the situation that traffic is crowded very much, target vehicle possibly overlap with other vehicle, and at this moment, traditional partitioning algorithm is difficult to be partitioned into the connected region that comprises single unit vehicle.In order to overcome this problem, adopt the method tracking target vehicle of following the tracks of vehicle characteristics point at present more and more, just need cut apart many vehicle characteristics point of adhesion this moment, follows the tracks of efficient to improve.
Traditional K-means clustering algorithm is not chosen initial cluster center with random approach, the difference of selected point, and cluster result maybe be just different, and such dependence just causes the instability of cluster result, and is absorbed in local optimum easily but not global optimum's cluster result; And this clustering algorithm is very sensitive to noise spot and isolated point; Cluster result depends on the setting of initial value, but the selected of k value (cluster number) often will just can find optimum value through experiment many times.Though a lot of improvement algorithms is arranged at present, how confirms correctly that initial cluster center and cluster number remain to need the technical matters that overcomes badly.
Summary of the invention
The objective of the invention is to have overcome above-mentioned shortcoming of the prior art, a kind of introducing area matching algorithm be provided, fixing cluster number, processing speed faster in the Video Detection based on the image segmentation disposal route of area matched optimization K-means clustering algorithm.
In order to realize above-mentioned purpose, the image segmentation disposal route based on area matched optimization K-means clustering algorithm in the Video Detection of the present invention may further comprise the steps:
(0) behind the extraction present frame movement destination image, finds the target image that is complementary with described movement destination image in the former frame through coupling;
(1) extracts present frame and former frame motion characteristics point respectively;
(2) present frame and former frame are carried out area matched, obtain the overlapping overlapping area zone of present frame movement destination image and former frame movement destination image;
(3) calculate the overlapping area zone of present frame and the unique point average of non-overlapped surface area respectively, as K-average initial cluster center;
(4) carry out the K-mean cluster and cut apart, and revise segmentation result;
(5) successfully whether judgement cut apart, if success then gets into step (8), if unsuccessful, then gets into step (6);
(6) relatively whether this segmentation result is consistent with a preceding segmentation result, if consistent, then gets into step (8), if inconsistent, then gets into step (7);
(7) adjust initial cluster center according to the result after cutting apart, and return step (4);
(8) output result, method ends.
Be somebody's turn to do in the image segmentation disposal route based on area matched optimization K-means clustering algorithm, described K-mean cluster is divided into the 2-mean cluster and cuts apart.
Be somebody's turn to do in the image segmentation disposal route based on area matched optimization K-means clustering algorithm, described unique point comprises area features point, position feature point and Feature Points.
Be somebody's turn to do in the image segmentation disposal route based on area matched optimization K-means clustering algorithm, described step (4) is specially: carry out the K-mean cluster and cut apart, and according to cluster result, revise the classification situation of eigenwert in the overlapping area zone.
In should the image segmentation disposal route based on area matched optimization K-means clustering algorithm, whether success be cut apart in the judgement in the described step (5), is specially: whether the area of judging the target image after the cluster according to the area of previous frame target image is reasonable.
Be somebody's turn to do in the image segmentation disposal route based on area matched optimization K-means clustering algorithm; Result after described basis is cut apart adjusts initial cluster center; Be specially: exceed with the cluster feature point that has target image according to the result of cutting apart (7), readjusts initial cluster center.
Be somebody's turn to do in the image segmentation disposal route based on area matched optimization K-means clustering algorithm, described movement destination image is a vehicle image.
Adopted in the Video Detection of the present invention image segmentation disposal route based on area matched optimization K-means clustering algorithm; The vehicle characteristics point of two field picture (comprised area before and after it at first extracted; The position; Shape), because camera shutter speed is enough fast, so front and back two frame vehicle areas exist overlapping certainly.The overlapping situation of area of two frame vehicles before and after the contrast, the position that can tentatively obtain following the tracks of the vehicle headstock or the tailstock.If back one frame vehicle and other automobile storage in adhesion, are extracted characteristic point position in the area overlapping region, ask for these unique point averages, extract residue character point position simultaneously and ask for its average.Two averages that obtain this moment are respectively as two types of initial cluster center points to be split.Carrying out the K-average then cuts apart; Classification situation according to unique point in the above-mentioned area of the cluster result correction overlapping region of output; Judge according to the previous frame vehicle area whether the vehicle after cluster this moment is reasonable simultaneously, if unreasonable, again to cluster result cluster again.Only cut apart that piece cluster feature crowd who has vehicle to be tracked this moment, add up cluster centre again according to above method simultaneously.After finding rational vehicle, finish cluster segmentation, the returning tracking result.Do not have difference if cut apart the back cluster result repeatedly, and do not find suitable cluster result, then get into next step matching treatment.This method is based on area matched optimization; And employing fixedly cluster number is cut apart; The vehicle target that obtains after process K-average is cut apart no longer need get into the matching treatment of next round, makes processing speed faster, has practiced thrift the processing time of whole video image target following.
Description of drawings
Fig. 1 is based on the process flow diagram of the image segmentation disposal route of area matched optimization K-means clustering algorithm in the Video Detection of the present invention.
Fig. 2 is the process flow diagram of the area matched optimization K-means clustering algorithm that the present invention adopted.
Fig. 3 is the synoptic diagram of feature extraction in the area matched optimization K-means clustering algorithm that the present invention adopted.
Fig. 4 extracts synoptic diagram for middle cluster centre in the area matched optimization K-means clustering algorithm that the present invention adopted.
Embodiment
In order more to be expressly understood technology contents of the present invention, the special following examples of lifting specify.
See also shown in Figure 1, in the Video Detection of the present invention based on the process flow diagram of a kind of embodiment of the image segmentation disposal route of area matched optimization K-means clustering algorithm.
In this embodiment, described method may further comprise the steps:
(0) behind the extraction present frame movement destination image, finds the target image that is complementary with described movement destination image in the former frame through coupling;
(1) extracts present frame and former frame motion characteristics point respectively;
(2) present frame and former frame are carried out area matched, obtain the overlapping overlapping area zone of present frame movement destination image and former frame movement destination image;
(3) calculate the overlapping area zone of present frame and the unique point average of non-overlapped surface area respectively, as K-average initial cluster center;
(4) carry out the K-mean cluster and cut apart, and revise segmentation result;
(5) successfully whether judgement cut apart, if success then gets into step (8), if unsuccessful, then gets into step (6);
(6) relatively whether this segmentation result is consistent with a preceding segmentation result, if consistent, then gets into step (8), if inconsistent, then gets into step (7);
(7) adjust initial cluster center according to the result after cutting apart, and return step (4);
(8) output result, method ends.
In this embodiment, described movement destination image is a vehicle image, and described K-mean cluster is divided into the 2-mean cluster to be cut apart, and described unique point comprises area features point, position feature point and Feature Points.
Of the present invention a kind of preferred embodiment in, the step (4) based on the image segmentation disposal route of area matched optimization K-means clustering algorithm in the described Video Detection is specially:
Carry out the K-mean cluster and cut apart, and, revise the classification situation of eigenwert in the overlapping area zone according to cluster result.
In another preferred embodiment of the present invention, cut apart whether success based on the judgement in the step (5) of the image segmentation disposal route of area matched optimization K-means clustering algorithm in the described Video Detection, be specially:
Whether the area of judging the target image after the cluster according to the area of previous frame target image is reasonable.
Result after described basis is cut apart adjusts initial cluster center, is specially:
(7) according to the result of cutting apart, exceed, readjust initial cluster center with the cluster feature point that has target image.
In practical application of the present invention, the flow process of the improved K-means clustering algorithm that is adopted is as shown in Figure 2, may further comprise the steps:
(1) behind the extraction present frame moving target, finds the former frame vehicle that is complementary with it through coupling;
(2) as shown in Figure 3, extract present frame and former frame motion characteristics point respectively;
(3) as shown in Figure 4, two vehicles are carried out area matched, obtain the overlapping area zone;
(4) as shown in Figure 4, calculate the unique point average of the present frame of overlapping area zone and non-overlapped surface area respectively, as K-average initial cluster center;
(5) carry out the 2-mean cluster and cut apart, revise segmentation result, whether judgement is cut apart successful; If unsuccessful, adjust initial cluster center according to the result after cutting apart,
(6) repeating step 5), up to cutting apart success or cutting apart convergence;
(7) output result.
Generally speaking, many vehicles phenomenon that sticks together is less relatively in the zone of need following the tracks of, and vehicle characteristics point is few, so the operand of this algorithm and not quite, processing speed is faster; Do not have difference if cut apart the back cluster result repeatedly, and do not find suitable cluster result, then get into next step matching treatment, in the case,, promptly can guarantee real-time follow-up yet though there is the repetition cutting procedure.
K-average dividing method after the improvement; It is based on area matched optimization; And employing fixedly cluster number is cut apart; The vehicle target that obtains after process K-average is cut apart no longer need get into the matching treatment of next round, makes processing speed faster, has practiced thrift the processing time of whole video image target following.
In this instructions, the present invention is described with reference to its certain embodiments.But, still can make various modifications and conversion obviously and not deviate from the spirit and scope of the present invention.Therefore, instructions and accompanying drawing are regarded in an illustrative, rather than a restrictive.
Claims (7)
- In the Video Detection based on the image segmentation disposal route of area matched optimization K-means clustering algorithm, it is characterized in that described method may further comprise the steps:(0) behind the extraction present frame movement destination image, finds the target image that is complementary with described movement destination image in the former frame through coupling;(1) extracts present frame and former frame motion characteristics point respectively;(2) present frame and former frame are carried out area matched, obtain the overlapping overlapping area zone of present frame movement destination image and former frame movement destination image;(3) calculate the overlapping area zone of present frame and the unique point average of non-overlapped surface area respectively, as K-average initial cluster center;(4) carry out the K-mean cluster and cut apart, and revise segmentation result;(5) successfully whether judgement cut apart, if success then gets into step (8), if unsuccessful, then gets into step (6);(6) relatively whether this segmentation result is consistent with a preceding segmentation result, if consistent, then gets into step (8), if inconsistent, then gets into step (7);(7) adjust initial cluster center according to the result after cutting apart, and return step (4);(8) output result, method ends.
- 2. based on the image segmentation disposal route of area matched optimization K-means clustering algorithm, it is characterized in that described K-mean cluster is divided into the 2-mean cluster and cuts apart in the Video Detection according to claim 1.
- 3. based on the image segmentation disposal route of area matched optimization K-means clustering algorithm, it is characterized in that described unique point comprises area features point, position feature point and Feature Points in the Video Detection according to claim 1.
- 4. based on the image segmentation disposal route of area matched optimization K-means clustering algorithm, it is characterized in that described step (4) is specially in the Video Detection according to claim 1:Carry out the K-mean cluster and cut apart, and, revise the classification situation of eigenwert in the overlapping area zone according to cluster result.
- 5. based on the image segmentation disposal route of area matched optimization K-means clustering algorithm, it is characterized in that in the Video Detection according to claim 1 that whether successful judgement in the described step (5) is cut apart, and is specially:Whether the area of judging the target image after the cluster according to the area of previous frame target image is reasonable.
- 6. based on the image segmentation disposal route of area matched optimization K-means clustering algorithm, it is characterized in that the result after described basis is cut apart adjusts initial cluster center, is specially in the Video Detection according to claim 1:(7) according to the result of cutting apart, exceed, readjust initial cluster center with the cluster feature point that has target image.
- According in each described Video Detection in the claim 1 to 6 based on the image segmentation disposal route of area matched optimization K-means clustering algorithm, it is characterized in that described movement destination image is a vehicle image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010254746.7A CN102375983B (en) | 2010-08-17 | 2010-08-17 | Based on the image segmentation processing method of area matching optimization K-means clustering algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010254746.7A CN102375983B (en) | 2010-08-17 | 2010-08-17 | Based on the image segmentation processing method of area matching optimization K-means clustering algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102375983A true CN102375983A (en) | 2012-03-14 |
| CN102375983B CN102375983B (en) | 2015-11-25 |
Family
ID=45794563
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201010254746.7A Active CN102375983B (en) | 2010-08-17 | 2010-08-17 | Based on the image segmentation processing method of area matching optimization K-means clustering algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102375983B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105930833A (en) * | 2016-05-19 | 2016-09-07 | 重庆邮电大学 | Vehicle tracking and segmenting method based on video monitoring |
| CN106570877A (en) * | 2016-10-27 | 2017-04-19 | 西安科技大学 | Coal mining machine pose positioning system and method based on coal mining machine virtual prototype and real image registration |
| CN109800684A (en) * | 2018-12-29 | 2019-05-24 | 上海依图网络科技有限公司 | The determination method and device of object in a kind of video |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1575014A2 (en) * | 2001-06-21 | 2005-09-14 | Fujitsu Limited | Method and apparatus for processing pictures of vehicles |
| CN101159855A (en) * | 2007-11-14 | 2008-04-09 | 南京优科漫科技有限公司 | Characteristic point analysis based multi-target separation predicting method |
-
2010
- 2010-08-17 CN CN201010254746.7A patent/CN102375983B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1575014A2 (en) * | 2001-06-21 | 2005-09-14 | Fujitsu Limited | Method and apparatus for processing pictures of vehicles |
| CN101159855A (en) * | 2007-11-14 | 2008-04-09 | 南京优科漫科技有限公司 | Characteristic point analysis based multi-target separation predicting method |
Non-Patent Citations (1)
| Title |
|---|
| 陈功，杨奎元，周荷琴，严捷丰: "鲁棒的实时多车辆检测与跟踪系统设计", 《信号处理》, vol. 25, no. 4, 30 April 2009 (2009-04-30) * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105930833A (en) * | 2016-05-19 | 2016-09-07 | 重庆邮电大学 | Vehicle tracking and segmenting method based on video monitoring |
| CN105930833B (en) * | 2016-05-19 | 2019-01-22 | 重庆邮电大学 | A kind of vehicle tracking and dividing method based on video monitoring |
| CN106570877A (en) * | 2016-10-27 | 2017-04-19 | 西安科技大学 | Coal mining machine pose positioning system and method based on coal mining machine virtual prototype and real image registration |
| CN109800684A (en) * | 2018-12-29 | 2019-05-24 | 上海依图网络科技有限公司 | The determination method and device of object in a kind of video |
| CN109800684B (en) * | 2018-12-29 | 2022-06-21 | 上海依图网络科技有限公司 | Method and device for determining object in video |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102375983B (en) | 2015-11-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Sivaraman et al. | A review of recent developments in vision-based vehicle detection | |
| Cao et al. | Multi-view frustum pointnet for object detection in autonomous driving | |
| CN104299008A (en) | Vehicle type classification method based on multi-feature fusion | |
| Huang et al. | Lane detection based on inverse perspective transformation and Kalman filter | |
| CN103903019A (en) | Automatic generating method for multi-lane vehicle track space-time diagram | |
| CN106919902B (en) | Vehicle identification and track tracking method based on CNN | |
| Chavez-Garcia et al. | Frontal object perception using radar and mono-vision | |
| EP3095073A1 (en) | Vehicle detection system and method thereof | |
| CN104537360A (en) | Method and system for detecting vehicle violation of not giving way | |
| CN102147861A (en) | Moving target detection method for carrying out Bayes judgment based on color-texture dual characteristic vectors | |
| WO2008139529A1 (en) | Vehicle periphery monitoring system, vehicle periphery monitoring program and vehicle periphery monitoring method | |
| CN103927763B (en) | A kind of image sequence multiple target tracking track identification processing method | |
| CN103345748A (en) | Positioning and partition method for human tissue cell two-photon microscopic image | |
| CN102375983A (en) | Image segmentation processing method based on area matching optimization K-means clustering algorithm | |
| US11845466B2 (en) | Normal distributions transform (NDT) method for LiDAR point cloud localization in unmanned driving | |
| CN102194129B (en) | Vehicle-type-clustering-based video detection method for traffic flow parameters | |
| WO2021013227A1 (en) | Image processing method and apparatus for target detection | |
| Guindel et al. | Joint object detection and viewpoint estimation using CNN features | |
| Cai et al. | Real-time arrow traffic light recognition system for intelligent vehicle | |
| CN103886609A (en) | Vehicle tracking method based on particle filtering and LBP features | |
| CN104408400B (en) | It is a kind of that multi-target detection method can not be differentiated based on single image frequency domain information | |
| Rakotondrajao et al. | Road boundary detection for straight lane lines using automatic inverse perspective mapping | |
| CN113221957B (en) | Method for enhancing radar information fusion characteristics based on center | |
| JP2023541967A (en) | Computer-implemented method for continuous adaptive detection of environmental features during automatic and assisted driving of own vehicle | |
| Sato et al. | Crossing obstacle detection with a vehicle-mounted camera |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |