CN102679957B - Background information and color feature combined fish body detection method - Google Patents
Background information and color feature combined fish body detection method Download PDFInfo
- Publication number
- CN102679957B CN102679957B CN 201210124965 CN201210124965A CN102679957B CN 102679957 B CN102679957 B CN 102679957B CN 201210124965 CN201210124965 CN 201210124965 CN 201210124965 A CN201210124965 A CN 201210124965A CN 102679957 B CN102679957 B CN 102679957B
- Authority
- CN
- China
- Prior art keywords
- image
- fish body
- background
- target
- body target
- 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.)
- Expired - Fee Related
Links
Images
Landscapes
- Image Analysis (AREA)
Abstract
The invention provides a background information and color feature combined fish body detection method. The method includes the steps that a computer reads a background image and an image which includes a fish body target; background subtraction is used for calculating the two images so as to obtain an initial fish body moving target, and a plurality of noise points and background information points are included in the images at the moment; in order to obtain the accurate fish body target, information about a red, green, blue (RGB) color model is used for obtaining the RGB color information of the fish body form an original color image, and the image is processed through an (R-B) operator; the processed image target presents a large difference from a background pixel, a histogram is in a bimodal distribution, the peak point of the histogram is found as a threshold value point, the image after the (R-B) processing is subjected to a threshold processing, the pixel larger than the threshold value is the fish body target, and other pixels are ejected; and the image output by the computer at the moment is the final detecting result of the fish body target. The fish body detection method has the advantages of being suitable for fish body detection in the scenes with simple backgrounds and fixed video cameras, good in generality, accurate in detecting target, high in speed, and the like.
Description
Technical field
The invention belongs to technical field of computer vision, especially a kind of relatively simple and fixing scene of video camera of background that is applicable to, carry out real-time detection and the high method of degree of accuracy to fish body target.
Background technology
In the economic technology high speed development, environmental pollution is inevitable, thereby causes greenhouse effect serious, global warming, and water body environment is deeply influenced especially, becomes the person that finally do not take in of pollution.Therefore, how effectively the water body environment quality to be carried out to monitoring management, the control water pollution has become the problem that countries in the world are paid close attention to, and actively drop into to pay close attention to Water quality and water body are carried out safely to reliable monitoring and early warning timely, the research that makes Water quality management work and water pollution prevent and treat method also highlights its importance increasingly.What the water body environment monitoring was generally applied is biological method, the biological monitoring technology is combined with environmental science.Home and abroad environment field of scientific study common concern at present is in utilizing the biological monitoring technology to set up the water body environment safety pre-warning system.The theoretical core of biological monitoring technology and water body environment early warning becomes possibility for utilizing behavior reaction to carry out directly monitoring rapidly, thereby the reflection water body environment changes the impact on the hydrobiont survival condition.Utilize some fish characteristics sensitive to the reacting condition of water body chemical composition, be applied to the research of pollutant component in water body, not only can obtain single action effect of planting pollutant in water-outlet body, also can reflect the composite pollution of Multiple components, can be used as the overall target of estimating water pollution, therefore, the variation that environment change causes Fish behavior to occur can be used as us and monitors the scientific basis of water pollution and good experiment material.Especially along with the development of computer vision technique, adopt, based on the video monitoring technology, hydrobiont is carried out to motion target tracking, change to determine biological behavior.The advantage of this method is the operating process simple and fast, can to the water body environment quality, be monitored in real time.Fish are biological monitoring one of most widely used hydrobionts in water body environment pollution research.Utilize Fish behavior monitoring water pollution, early, Belding is according to the respiratory variations indication toxic environment of fish for the application of the states such as America and Europe; Wxlde reflects the impact of waste water of paper mill on fish by the cough number of times of observing fish, and the safe concentration of definite waste water; A kind of mosquito fish test wastewater toxicity of small volume for Davis, subsequently, the swimming behavior of fish, positive rheotaxis and selection behavior etc. also are applied in succession.China's research is in this respect started late, the pollution condition of organic agricultural chemicals in the noisy cholinesterase activity monitoring water body of the aquatic fish used of the Chinese Academy of Sciences; Qin Dongli etc. utilize fish to monitor and estimate the degree of mercury pollution in water body environment; Wang Hongjun, Liu Shuying etc. utilize the susceptibility of fish breathing function aspect, according to the variation of the respiration parameters of fish, carry out the early warning water body environment pollution.Existing all researchs are all to be based upon on the basis of fish target detection, how can detect quickly and accurately the bottleneck that fish body target becomes every research.
In monitoring method used, it is good that the background subtraction method has real-time, and the characteristics that computation complexity is low can be cut apart fish body target well in the situation that environment is more satisfactory.But in the situation that ripples and illumination, noise etc. disturb, owing to can not effectively removing noise and unnecessary background dot, thereby there is the complete or not too much phenomenon of target detection, make fish body target detection effect poor.And color characteristic information is based on the method for a kind of relative flexibility of color model, implementation procedure is simple, calculated amount is few.
The phenomenon that exists rapidity and accuracy mutually to restrict in fish body target detection, obtain accurate fish body target just needs through complicated algorithm if want, and complicated algorithm will make calculated amount large, cause computing time long, thereby rapidity is also just poor.Improve rapidity if want and just must reduce calculated amount, and calculated amount one reduces, the degree of accuracy of algorithm just can not guarantee.
Summary of the invention
In order to solve the problem that rapidity and accuracy restrict mutually in the fish target detection, the present invention is based on the minimum background subtraction method of calculated amount, simultaneously, in order to overcome the impact on fish body target detection precision of factors such as improper degree that external condition chooses as illumination, noise and background, merge color characteristic and carried out threshold process on the basis of background subtraction method, can detect fast and accurately fish body target.
The phenomenon mutually restricted in order to overcome rapidity in fish body target detection and accuracy, the present invention proposes a kind of new fusion background information and the fish body detecting method of color characteristic, can effectively solve the problem that this type of restricts mutually.
In order to solve the technical matters of above-mentioned existence, the present invention adopts following technical proposals: a kind of fish body detecting method that merges background information and color characteristic comprises the following steps:
(1) background obtains
After video camera is fixing, take the image image input as a setting computing machine while there is no fish body target;
(2) background subtraction method
Take fish body target image, with background image, utilize the background subtraction method to obtain fish body target Preliminary detection image, then setting threshold is differentiated, thereby obtains the bianry image of fish body target Preliminary detection;
(3) obtain the colouring information of fish body target
Utilize the RGB colored pixels information reverting bianry image of the fish body target image photographed in step (2), thereby obtain the colouring information image of fish body target;
(4) select suitable combination operator
Contain R, G, tri-different components of B in the RGB color model of fish volume image, select
as the computing operator;
(5) the fish volume image is carried out to threshold test
The image that utilizes (R-B) operator to obtain step (3) carries out computing, and image object and background pixel after processing present larger difference, and its histogram is bimodal distribution, and finding histogrammic peak valley point is threshold point T;
(6) threshold determination
The image that step (5) is obtained carries out threshold process, and the pixel that is greater than threshold value T is fish body target, otherwise rejects; Now the image of computer export is final fish body target detection result.
Background subtraction method real-time is good, to the registration of target, can well provide the information such as position, size, shape of moving target, just the key issue of the method be how to obtain accurately, background image reliably.The acquisition methods of background image specifically has following two kinds: a kind of is in ideal conditions, in the situation that the image that scene does not have moving target to obtain image as a setting, then with each frame in sequence of video images therewith background image carry out calculus of differences, obtain moving target.The second is in the situation that real, because the actual life Scene at every moment all changes, detection for moving target, other the interference of the external world such as illumination, weather all can affect final testing result, the adverse effect of bringing in order to reduce these variations, need to set up background model and background model real-time update.
The present invention is applicable to the condition that scene is simple and video camera is static, so adopt first method to obtain background image, then with each frame in sequence of video images therewith background image carry out calculus of differences, obtain moving target, thus minimizing calculated amount and time.
Computing formula is as follows:
Wherein,
for the two field picture in video sequence,
for background image,
it is the target area obtained after two two field picture difference.
In order to reach the accurate purpose of obtaining moving target lacked of energy, further by the method for Threshold segmentation, target area to be processed, formula is as follows:
Wherein,
for carrying out the threshold value of binary conversion treatment, generally adopt the adaptive approach selected threshold.The quality that this threshold value is chosen, determining can be complete the moving target of choosing, the elimination noise that simultaneously it can also be suitable.
On the basis of background subtraction method, the Fusion of Color feature is carried out the threshold value secondary splitting, solved impact fish body target detection precision caused due to illumination, ripples shade, noise etc., also solved the target over-segmentation of fish body and the less divided phenomenon occurred due to the background subtraction method, thereby can obtain accurately fish body target simultaneously.
What color characteristic adopted is the RGB model, and according to the principle of three primary colours, in the RGB color model, any coloured light F can mix with the addition of R, G, tri-different components of B:
.The Selection and Constitute operator
carry out algebraic operation as characteristic quantity, the image obtained after computing shows that the pixel value of fish body target and background has larger difference, its histogram becomes obvious bimodal distribution, finds the threshold value of peak valley point as secondary splitting, just can obtain fast accurately fish body target.
Owing to adopting technique scheme, the fish body detecting method of fusion background information provided by the invention and color characteristic has such beneficial effect: the present invention adopts the minimum background subtraction method of calculated amount to obtain fish body target primary detection result, greatly shortened the target detection time, there is good rapidity, and on the basis of background subtraction method, utilized color characteristic information to carry out secondary splitting to the fish volume image, solved in the background subtraction method and unnecessary noise spot and the phenomenon of background dot have easily occurred, color characteristic information calculated amount is also very little simultaneously, so the quick performance of fish body target detection accesses assurance, effect can be found out by experiment, fish health check-up measuring tool in conjunction with two kinds of algorithms has versatility good, the detection target is accurate, the characteristics such as speed is fast.
The accompanying drawing explanation
Fig. 1 is the fish body detecting method process flow diagram that background information and color characteristic are merged in the present invention;
Fig. 2 is the experimental result picture of background subtraction method;
Fig. 3 is (R-B) chromaticity difference diagram;
Fig. 4 is the fish body detecting method experimental result picture that background information and color characteristic are merged in the present invention.
Embodiment
Below in conjunction with accompanying drawing, the specific embodiment of the present invention is described in further detail.
The fusion background information that the present invention proposes and the fish body detecting method of color characteristic as shown in Figure 1, below are divided into six parts and introduce embodiment.
(1) background obtains
After video camera is fixing, take the image image input as a setting computing machine while there is no fish body target.
(2) background subtraction method
Take fish body target image, with background image, utilize formula (1) to do the poor fish body target Preliminary detection image that obtains, because background subtraction method calculated amount is little, operation time is short, can obtain fast target Preliminary detection image, but now image comprises noise spot and background dot, need to carry out the setting threshold differentiation, thereby obtain the bianry image of fish body target Preliminary detection, as shown in Figure 2.In order to obtain more accurately fish body target, must further process.
(3) obtain the colouring information of fish body target
Utilize the fish body target bianry image in the RGB colored pixels information reverting step (2) of the fish body target image photographed in step (2), thereby obtain the colouring information image of fish body target.
(4) select suitable combination operator
Contain R, G, tri-different components of B in the RGB color model of fish volume image, when cutting apart, the Selection and Constitute operator
,
,
and
carry out algebraic operation as characteristic quantity, found that fish body target
the pixel value of pixel value and background has larger difference, and the differentiation effect of other several operators is poor, so, select
as the computing operator.
(5) the fish volume image is carried out to threshold test
The image that utilizes (R-B) operator to obtain step (3) carries out computing, and figure as shown in Figure 3 as a result; Image object and background pixel after processing present larger difference, and its histogram is bimodal distribution, and finding histogrammic peak valley point is threshold point T.
(6) threshold determination
The image 3 that step (5) is obtained carries out threshold process, and the pixel that is greater than threshold value T is fish body target, otherwise rejects; Now the image of computer export is final fish body target detection result.As shown in Figure 4.As can be seen from Figure 4, this method can accurately obtain fish body target.
Claims (1)
1. merge the fish body detecting method of background information and color characteristic, this detection method content comprises:
(1) background obtains
After video camera is fixing, take the image image input as a setting computing machine while there is no fish body target;
(2) background subtraction method
Take fish body target image, with background image, utilize the background subtraction method to obtain fish body target Preliminary detection image, then setting threshold is differentiated, thereby obtains the bianry image of fish body target Preliminary detection;
It is characterized in that: this detection method content also comprises the steps:
(3) obtain the colouring information of fish body target
Utilize the RGB colored pixels information reverting bianry image of the fish body target image photographed in step (2), thereby obtain the colouring information image of fish body target;
(4) select suitable combination operator
Contain R, G, tri-different components of B in the RGB color model of fish volume image, select
as the computing operator;
(5) the fish volume image is carried out to threshold test
The image that utilizes (R-B) operator to obtain step (3) carries out computing, and image object and background pixel after processing present larger difference, and its histogram is bimodal distribution, and finding histogrammic peak valley point is threshold point T;
(6) threshold determination
The image that step (5) is obtained carries out threshold process, and the pixel that is greater than threshold value T is fish body target, otherwise rejects; Now the image of computer export is final fish body target detection result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201210124965 CN102679957B (en) | 2012-04-26 | 2012-04-26 | Background information and color feature combined fish body detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201210124965 CN102679957B (en) | 2012-04-26 | 2012-04-26 | Background information and color feature combined fish body detection method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102679957A CN102679957A (en) | 2012-09-19 |
| CN102679957B true CN102679957B (en) | 2013-12-18 |
Family
ID=46812244
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 201210124965 Expired - Fee Related CN102679957B (en) | 2012-04-26 | 2012-04-26 | Background information and color feature combined fish body detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102679957B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104749334A (en) * | 2015-02-16 | 2015-07-01 | 燕山大学 | Mode-recognition-based design method for biological abnormal water quality evaluation system |
| CN108288388A (en) * | 2018-01-30 | 2018-07-17 | 深圳源广安智能科技有限公司 | A kind of intelligent traffic monitoring system |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR100224729B1 (en) * | 1996-12-10 | 1999-10-15 | 윤종용 | Ferroelectric capacitor for semiconductor device and fabricating method thereof |
| JP2001116527A (en) * | 1999-10-20 | 2001-04-27 | Matsushita Electric Ind Co Ltd | Method and device for detecting solid body |
| TW201002073A (en) * | 2008-06-23 | 2010-01-01 | Huper Lab Co Ltd | Method of vehicle segmentation and counting for nighttime video frames |
-
2012
- 2012-04-26 CN CN 201210124965 patent/CN102679957B/en not_active Expired - Fee Related
Non-Patent Citations (7)
| Title |
|---|
| 《动态图像序列中鱼体体干运动建模》;程淑红 等;《光电工程》;20120331;第39卷(第3期);125-129页 * |
| 《基于视频监测的鱼类运动跟踪研究》;蔡菁;《硕士毕业论文》;20110831;1-62 * |
| 《彩色图像分割方法综述》;林开颜 等;《中国图象图形学报》;20050131;第10卷(第1期);1-10 * |
| JP特开2001-116527A 2001.04.27 |
| 林开颜 等.《彩色图像分割方法综述》.《中国图象图形学报》.2005,第10卷(第1期),1-10. |
| 程淑红 等.《动态图像序列中鱼体体干运动建模》.《光电工程》.2012,第39卷(第3期),125-129页. |
| 蔡菁.《基于视频监测的鱼类运动跟踪研究》.《硕士毕业论文》.2011,1-62. |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102679957A (en) | 2012-09-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107478340B (en) | A kind of converter valve monitoring method and system based on image co-registration | |
| CN104680556A (en) | Parallax-based three-dimensional trajectory tracking method of fish movement | |
| CN102131101A (en) | Intelligent video image quality automatic analysis system and method | |
| CN103149214B (en) | Method for detecting flaw on surface of fruit | |
| CN103679146A (en) | Safety belt intelligent detection method based on high-pass filter and Hough conversion | |
| CN102765249A (en) | Machine vision detection method based on four-colour printed matter registration detection marks | |
| CN103439348A (en) | Remote controller key defect detection method based on difference image method | |
| CN102679957B (en) | Background information and color feature combined fish body detection method | |
| CN105260716A (en) | Fault indicator state identification method and fault indicator state identification device | |
| CN103810696A (en) | Method for detecting image of target object and device thereof | |
| CN103971479B (en) | Forest fires localization method based on camera calibration technology | |
| CN113095441A (en) | Pig herd bundling detection method, device, equipment and readable storage medium | |
| CN113191270B (en) | Method and device for detecting throwing event, electronic equipment and storage medium | |
| CN207366475U (en) | A kind of relief pattern Online Video detection device | |
| CN105092589A (en) | Detection method for defects of capsule head | |
| Mao et al. | Design of visual navigation system of farmland tracked robot based on raspberry pie | |
| CN112818836B (en) | Method and system for detecting personnel target of transformer substation scene | |
| CN115437394A (en) | Unmanned aerial vehicle system for river channel monitoring and autonomous line patrol method thereof | |
| US20140055462A1 (en) | Method of defining image surveillance region | |
| Junlong et al. | Helmet Detection Algorithm Based on the Improved YOLOv5 and Dynamic Anchor Box Matching | |
| Xiang et al. | A new model for daytime visibility index estimation fused average sobel gradient and dark channel ratio | |
| CN103909881B (en) | Environment information detecting system and environment information method for detecting | |
| CN111080677A (en) | Protection method for real-time partition operation of workers in pollution remediation site | |
| CN105959639B (en) | Pedestrian's monitoring method in avenue region based on ground calibration | |
| Fei et al. | Development of an Intelligent Monitoring System for Agricultural Machinery |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20131218 Termination date: 20160426 |