CN114022753A - Algorithm for detecting small empty targets based on significance and edge analysis - Google Patents
Algorithm for detecting small empty targets based on significance and edge analysis Download PDFInfo
- Publication number
- CN114022753A CN114022753A CN202111352007.6A CN202111352007A CN114022753A CN 114022753 A CN114022753 A CN 114022753A CN 202111352007 A CN202111352007 A CN 202111352007A CN 114022753 A CN114022753 A CN 114022753A
- Authority
- CN
- China
- Prior art keywords
- target
- unmanned aerial
- aerial vehicle
- algorithm
- image
- 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
- 238000004458 analytical method Methods 0.000 title claims abstract description 11
- 238000012216 screening Methods 0.000 claims abstract description 9
- 238000001514 detection method Methods 0.000 claims abstract description 8
- 230000007547 defect Effects 0.000 claims abstract description 5
- 238000000034 method Methods 0.000 claims description 14
- 230000011218 segmentation Effects 0.000 claims description 8
- 238000012545 processing Methods 0.000 claims description 5
- 238000010606 normalization Methods 0.000 claims description 3
- 238000003384 imaging method Methods 0.000 claims description 2
- 230000010339 dilation Effects 0.000 claims 1
- 238000013461 design Methods 0.000 description 4
- 238000003708 edge detection Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention provides a detection algorithm for a small blank target based on significance and edge analysis, which comprises the following steps of firstly, carrying out multi-scale Gaussian blur on an input image, and extracting significant target features according to the obvious features of the small-scale blurred target and the features of a large-scale blurred target with a large amount of neighborhood information; secondly, binarizing the salient target features according to the global feature distribution condition; expanding the binarization result, and connecting stray features generated by isolated dead points, noise and multi-texture background; fourthly, size screening is carried out on the expansion result, and the target position of the unmanned aerial vehicle is given by combining the characteristic strength and the height information of the target; through the steps, the algorithm filters interference caused by isolated dead points, noise points and complex backgrounds, and has strong robustness. The system changes the defects that the hidden danger of the unmanned aerial vehicle needs to be manually checked and people need to participate in real time in the traditional airport, and avoids the event that the unmanned aerial vehicle invades the airport.
Description
Technical Field
The invention designs a detection algorithm for small targets in air based on significance and edge analysis, and belongs to the technical field of video processing.
Background
Along with the development of low-cost, miniaturized unmanned aerial vehicle industry, unmanned aerial vehicle gets into civilian consumption field rapidly, spends several thousand yuan and can buy a fine performance unmanned aerial vehicle. However, at present, unauthorized unmanned aerial vehicles are used in large quantities, and hidden dangers are brought to public safety. How to discover the airport accessory unmanned aerial vehicle automatically becomes an important problem of the public safety guarantee at present.
The to-air small target detection algorithm that this patent designed has fused a plurality of strategies such as saliency, edge analysis, possesses the automated inspection ability to small target, low contrast target, has changed the airport in the tradition and needs artifical investigation unmanned aerial vehicle hidden danger, needs the drawback of people's real-time participation, avoids unmanned aerial vehicle to invade airport incident.
Disclosure of Invention
Object of the Invention
The invention aims to design an algorithm for detecting the small targets, the system integrates a plurality of strategies such as significance, edge analysis and the like, has automatic detection capability aiming at the small targets and the low-contrast targets, overcomes the defects that the hidden danger of an unmanned aerial vehicle needs to be manually checked and people need to participate in the airport in real time in the traditional airport, and avoids the event that the unmanned aerial vehicle invades the airport.
Technical scheme
The invention designs a detection algorithm for a small empty target based on significance and edge detection, which comprises the following processing flows:
step one, carrying out multi-scale Gaussian blur on an input image, and extracting the characteristic of a significant target according to the characteristic that a small-scale blurred target is obvious in characteristic and a large-scale blurred target has a large amount of neighborhood information
Secondly, binaryzation is carried out on the salient target features according to the global feature distribution condition
And step three, expanding the binarization result, and connecting stray features generated by isolated dead points, noise and multi-texture background.
And step four, size screening is carried out on the expansion result, and the target position of the unmanned aerial vehicle is given by combining the characteristic strength and the height information of the target.
Through the steps, the automatic detection of the unmanned aerial vehicle target in the image is realized by combining the saliency and the edge information of the target in the image. The system expands the image characteristic isolated points, connects multi-texture background objects, filters the objects according to the target size, and can effectively eliminate the interference of trees and buildings in common complex scenes. Through reserving the interface, the parameter in the method is convenient to adjust, and the method is suitable for imaging of each focal length lens, so that the flexibility of the system is improved. The algorithm changes the defects that the hidden danger of the unmanned aerial vehicle needs to be manually checked and people need to participate in real time in the traditional airport, and avoids the event that the unmanned aerial vehicle invades the airport.
In the step one, the input image is subjected to multi-scale gaussian blurring, which is performed as follows: and performing small-range Gaussian blur on the image to extract detail texture information, and simultaneously using large-range mean blur on the original image to extract large-range neighborhood information. And (4) taking a pixel-level Euclidean distance between the two to extract the significant features of the target.
Wherein, the binarization is performed according to the global feature distribution condition in the step two, and the method comprises the following steps: and (4) performing histogram statistics on the significant feature map obtained in the step one, taking the maximum value of the inter-class variance as a segmentation threshold, and performing binary segmentation on the feature map to suppress the interference of isolated dead points and noise points in the feature map.
Wherein, the method of expanding the binarization result in the step three is as follows: considering that the binary image obtained in the second step contains interference caused by complex background objects such as trees, buildings and the like, the binary image is expanded in a proper range to be connected with the stray characteristics of the complex multi-texture background.
Wherein, the step four includes the steps of performing size screening and giving the target position of the unmanned aerial vehicle by combining the target characteristic strength and the height information, and the method comprises the following steps: the expansion result obtained in step three eliminates the interference of isolated dead spots and noise points and connects the stray characteristics of the complex background together. Therefore, the connected domain search is carried out on the obtained expansion map in the step four, and the size of the target predicted size is presumed according to the current lens focal length, so as to guide the size screening of the connected domain. Then, the feature strength of the screened connected domain in the step one is scored, score normalization is carried out by combining height information, and if the highest score connected domain is larger than a preset value, an unmanned aerial vehicle target return value is output; otherwise, the unmanned aerial vehicle target does not exist in the current image.
Advantages of the invention
The invention has the advantages that the system can filter the interference caused by isolated dead points, noise points and complex backgrounds, and has strong robustness; the system reserves an interface, can adjust the size of an expected target according to the focal length of a camera lens, and has strong adaptability. The algorithm changes the defects that the hidden danger of the unmanned aerial vehicle needs to be manually checked and people need to participate in real time in the traditional airport, and avoids the event that the unmanned aerial vehicle invades the airport.
Drawings
Fig. 1 is an operation flow chart of an automatic unmanned aerial vehicle monitoring system.
Detailed Description
The invention designs a detection algorithm for a small empty target integrating significance and edge detection, the processing flow of the algorithm is shown in figure 1, and the specific processing steps are as follows:
step one, carrying out multi-scale Gaussian blur on an input image, and extracting the characteristic of a significant target according to the characteristic that a small-scale blurred target is obvious in characteristic and a large-scale blurred target has a large amount of neighborhood information
Secondly, binaryzation is carried out on the salient target features according to the global feature distribution condition
And step three, expanding the binarization result, and connecting stray features generated by isolated dead points, noise and multi-texture background.
And step four, size screening is carried out on the expansion result, and the target position of the unmanned aerial vehicle is given by combining the characteristic strength and the height information of the target.
Wherein the method of the step one is as follows:
and performing small-range Gaussian blur on the image to extract detail texture information, and simultaneously using large-range mean blur on the original image to extract large-range neighborhood information. And (4) taking a pixel-level Euclidean distance between the two to extract the significant features of the target. The pixel-level Euclidean distance calculation method comprises the following steps:
(1) first, an input image is acquired, which is set as IinThen, Iin(x, y) is the pixel value of the position (x, y) image, given a small scale Gaussian blur transform to fSLarge scale mean fuzzy transformation to fL。
(2) Calculating the blurred images of the size and the scale: f. ofS(Iin(x, y)) and fL(Iin(x,y))。
(3) The image salient feature map is calculated using the following formula:
Iout(x,y)＝(fs(Iin(x,y))-fL(Iin(x,y)))2
(4) the saliency map is normalized by the maximum and minimum values such that the output image range is [0,255 ].
Wherein the method of the second step is as follows:
and (4) performing histogram statistics on the significant feature map obtained in the step one, taking the maximum value of the inter-class variance as a segmentation threshold, and performing binary segmentation on the feature map to suppress the interference of isolated dead points and noise points in the feature map. The method for calculating the between-class variance is as follows:
(1) calculating the gray value accumulated value p of the gray level k (0-255)kCumulative mean mkAnd image global mean mG：
(2) Calculating the inter-class variance of the gray level k:
(3) taking the maximum value of the inter-class variance as a segmentation threshold eta:
and (4) segmenting the feature map by using a segmentation threshold eta, and binarizing the feature map into a binary map consisting of 0 and 255.
The method of the third step is as follows:
considering that the binary image obtained in the second step contains interference caused by complex background objects such as trees, buildings and the like, the binary image is expanded in a proper range to be connected with the stray characteristics of the complex multi-texture background.
Wherein the method of the fourth step is as follows:
the expansion result obtained in step three eliminates the interference of isolated dead spots and noise points and connects the stray characteristics of the complex background together. Therefore, the connected domain search is carried out on the obtained expansion map in the step four, and the size of the target predicted size is presumed according to the current lens focal length, so as to guide the size screening of the connected domain. Then, the feature strength of the screened connected domain in the step one is scored, score normalization is carried out by combining height information, and if the highest score connected domain is larger than a preset value, an unmanned aerial vehicle target return value is output; otherwise, the unmanned aerial vehicle target does not exist in the current image. The method comprises the following specific steps:
(1) let IinTraversing connected domains in the image for the inflation map:
C＝Contours(Iin)
(2) traversing the obtained expansion map, and filtering the size:
C′＝{Area(Ck)∈[Amin,Amax]|Ck∈C}
(3) let the characteristic diagram obtained in step one be IoutAnd scoring the screened connected domains according to the characteristic intensity:
(4) the scores are normalized in combination with the height information:
score′k＝scorek×log(y′k)
and (4) judging the threshold value of the normalized fraction to obtain whether the unmanned aerial vehicle and the position information of the unmanned aerial vehicle exist in the view field.
Claims (5)
1. An algorithm for detecting a small empty target based on significance and edge analysis is characterized by comprising the following processing flows:
step one, carrying out multi-scale Gaussian blur on an input image, and extracting the characteristic of a significant target according to the characteristic that a small-scale blurred target is obvious in characteristic and a large-scale blurred target has a large amount of neighborhood information
Secondly, binaryzation is carried out on the salient target features according to the global feature distribution condition
And step three, expanding the binarization result, and connecting stray features generated by isolated dead points, noise and multi-texture background.
And step four, size screening is carried out on the expansion result, and the target position of the unmanned aerial vehicle is given by combining the characteristic strength and the height information of the target.
Through the steps, the automatic detection of the unmanned aerial vehicle target in the image is realized by combining the saliency and the edge information of the target in the image. The algorithm expands the image characteristic isolated points, connects multi-texture background objects, filters the objects according to the target size, and can effectively eliminate the interference of trees and buildings in common complex scenes. Through reserving the interface, the parameter in the method is conveniently adjusted, the method is suitable for imaging of each focal length lens, and the flexibility of the algorithm is improved. The algorithm changes the defects that the hidden danger of the unmanned aerial vehicle needs to be manually checked and people need to participate in real time in the traditional airport, and avoids the event that the unmanned aerial vehicle invades the airport.
2. The algorithm for detecting the empty small target based on the significance and the edge analysis as claimed in claim 1, wherein:
the "multi-scale gaussian blurring of the input image" described in "step one" is performed as follows: and performing small-range Gaussian blur on the image to extract detail texture information, and simultaneously using large-range mean blur on the original image to extract large-range neighborhood information. And (4) taking a pixel-level Euclidean distance between the two to extract the significant features of the target.
3. The algorithm for detecting the empty small target based on the significance and the edge analysis as claimed in claim 1, wherein:
the "binarization according to the global feature distribution" in the "step two" is performed as follows: and (4) performing histogram statistics on the significant feature map obtained in the step one, taking the maximum value of the inter-class variance as a segmentation threshold, and performing binary segmentation on the feature map to suppress the interference of isolated dead points and noise points in the feature map.
4. The algorithm for detecting the empty small target based on the significance and the edge analysis as claimed in claim 1, wherein:
the "dilation of binarization results" described in "step three" is done as follows: considering that the binary image obtained in the second step contains interference caused by complex background objects such as trees, buildings and the like, the binary image is expanded in a proper range to be connected with the stray characteristics of the complex multi-texture background.
5. The algorithm for detecting the empty small target based on the significance and the edge analysis as claimed in claim 1, wherein:
the step four includes that size screening is carried out, and the target position of the unmanned aerial vehicle is given by combining target characteristic strength and height information, and the method comprises the following steps: the expansion result obtained in step three eliminates the interference of isolated dead spots and noise points and connects the stray characteristics of the complex background together. Therefore, the connected domain search is carried out on the obtained expansion map in the step four, and the size of the target predicted size is presumed according to the current lens focal length, so as to guide the size screening of the connected domain. Then, the feature strength of the screened connected domain in the step one is scored, score normalization is carried out by combining height information, and if the highest score connected domain is larger than a preset value, an unmanned aerial vehicle target return value is output; otherwise, the unmanned aerial vehicle target does not exist in the current image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111352007.6A CN114022753B (en) | 2021-11-16 | 2021-11-16 | Significance and edge analysis-based empty small target detection algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111352007.6A CN114022753B (en) | 2021-11-16 | 2021-11-16 | Significance and edge analysis-based empty small target detection algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114022753A true CN114022753A (en) | 2022-02-08 |
| CN114022753B CN114022753B (en) | 2024-05-14 |
Family
ID=80064313
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111352007.6A Active CN114022753B (en) | 2021-11-16 | 2021-11-16 | Significance and edge analysis-based empty small target detection algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114022753B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115294478A (en) * | 2022-07-28 | 2022-11-04 | 北京航空航天大学 | Aerial unmanned aerial vehicle target detection method applied to modern photoelectric platform |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103996209A (en) * | 2014-05-21 | 2014-08-20 | 北京航空航天大学 | Infrared vessel object segmentation method based on salient region detection |
| WO2016076449A1 (en) * | 2014-11-11 | 2016-05-19 | Movon Corporation | Method and system for detecting an approaching obstacle based on image recognition |
| CN109325935A (en) * | 2018-07-24 | 2019-02-12 | 国网浙江省电力有限公司杭州供电公司 | A kind of transmission line faultlocating method based on unmanned plane image |
| WO2020211522A1 (en) * | 2019-04-15 | 2020-10-22 | 京东方科技集团股份有限公司 | Method and device for detecting salient area of image |
-
2021
- 2021-11-16 CN CN202111352007.6A patent/CN114022753B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103996209A (en) * | 2014-05-21 | 2014-08-20 | 北京航空航天大学 | Infrared vessel object segmentation method based on salient region detection |
| WO2016076449A1 (en) * | 2014-11-11 | 2016-05-19 | Movon Corporation | Method and system for detecting an approaching obstacle based on image recognition |
| CN109325935A (en) * | 2018-07-24 | 2019-02-12 | 国网浙江省电力有限公司杭州供电公司 | A kind of transmission line faultlocating method based on unmanned plane image |
| WO2020211522A1 (en) * | 2019-04-15 | 2020-10-22 | 京东方科技集团股份有限公司 | Method and device for detecting salient area of image |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115294478A (en) * | 2022-07-28 | 2022-11-04 | 北京航空航天大学 | Aerial unmanned aerial vehicle target detection method applied to modern photoelectric platform |
| CN115294478B (en) * | 2022-07-28 | 2024-04-05 | 北京航空航天大学 | Aerial unmanned aerial vehicle target detection method applied to modern photoelectric platform |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114022753B (en) | 2024-05-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111292305B (en) | Improved YOLO-V3 metal processing surface defect detection method | |
| Al-Ghaili et al. | Vertical-edge-based car-license-plate detection method | |
| Gao et al. | Car license plates detection from complex scene | |
| CN110866903B (en) | Ping-pong ball identification method based on Hough circle transformation technology | |
| Paunwala et al. | A novel multiple license plate extraction technique for complex background in Indian traffic conditions | |
| Li et al. | Research on vehicle license plate location based on neural networks | |
| CN108563979B (en) | Method for judging rice blast disease conditions based on aerial farmland images | |
| Khalifa et al. | Malaysian Vehicle License Plate Recognition. | |
| CN109460722B (en) | Intelligent license plate recognition method | |
| CN111507145A (en) | Method, system and device for detecting barrier at storage position of embedded vehicle-mounted all-round looking system | |
| CN113409267B (en) | Pavement crack detection and segmentation method based on deep learning | |
| CN111027544B (en) | MSER license plate positioning method and system based on visual saliency detection | |
| Babbar et al. | A new approach for vehicle number plate detection | |
| CN111695373B (en) | Zebra stripes positioning method, system, medium and equipment | |
| CN109544535B (en) | Peeping camera detection method and system based on optical filtering characteristics of infrared cut-off filter | |
| CN112883986A (en) | Static infrared target lamp identification method under complex background | |
| CN115100104A (en) | Defect detection method, device and equipment for glass ink area and readable storage medium | |
| CN114359538A (en) | Water meter reading positioning and identifying method | |
| CN112288682A (en) | Electric power equipment defect positioning method based on image registration | |
| CN114022753A (en) | Algorithm for detecting small empty targets based on significance and edge analysis | |
| Miller et al. | Person tracking in UAV video | |
| Dhar et al. | Bangladeshi license plate recognition using adaboost classifier | |
| CN113657250A (en) | Flame detection method and system based on monitoring video | |
| CN112785550B (en) | Image quality value determining method and device, storage medium and electronic device | |
| CN110321828B (en) | Front vehicle detection method based on binocular camera and vehicle bottom shadow |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |