CN113160118A - Wire target detection method - Google Patents
Wire target detection method Download PDFInfo
- Publication number
- CN113160118A CN113160118A CN202110158342.6A CN202110158342A CN113160118A CN 113160118 A CN113160118 A CN 113160118A CN 202110158342 A CN202110158342 A CN 202110158342A CN 113160118 A CN113160118 A CN 113160118A
- Authority
- CN
- China
- Prior art keywords
- wire
- effective
- groups
- targets
- 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.)
- Pending
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 52
- 238000007781 pre-processing Methods 0.000 claims abstract description 10
- 238000000034 method Methods 0.000 claims description 19
- 238000001914 filtration Methods 0.000 claims description 9
- 230000002146 bilateral effect Effects 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 2
- 238000012795 verification Methods 0.000 claims description 2
- 238000004364 calculation method Methods 0.000 abstract description 7
- 238000003672 processing method Methods 0.000 abstract description 2
- 230000004927 fusion Effects 0.000 description 5
- 230000000007 visual effect Effects 0.000 description 4
- 238000006243 chemical reaction Methods 0.000 description 3
- 230000007547 defect Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 238000010276 construction Methods 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
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
Landscapes
- Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
- Length Measuring Devices By Optical Means (AREA)
Abstract
The invention discloses a wire target detection method, which comprises the following steps: (1) acquiring original depth data of a scene where the electric wire is located through a laser sensor, and converting the original depth data into a gray image according to the distance of the depth; (2) carrying out data preprocessing on the gray level image obtained in the step (1) to obtain a binary image; (3) obtaining linear targets in all directions by using Hough transform on the binary image obtained in the step (2) and taking the linear targets as reference wire targets; (4) converting the reference wire target obtained in the step (3) into an intercept-slope characteristic space, and denoising according to the slope and the intercept respectively to obtain an effective reference wire target; (5) and (5) connecting the effective reference wire target obtained in the step (4) as a final wire target to obtain a wire target detection result. The invention only needs depth data, and effectively filters noise in the depth data by using a digital image processing method, and simultaneously improves the calculation efficiency and the accuracy of the wire detection result.
Description
Technical Field
The invention relates to the technical field of target detection, in particular to a wire target detection method.
Background
Traditional power line patrols and examines, the overlap joint flow need rely on the electric power personnel operation in person, consequently, can endanger line patrol workman's life safety and danger in dangerous section, like high altitude construction. The invention relates to an electric power robot, in particular to an electric operating robot, which solves the safety problem that the electric power industry depends on manual operation. The live working robot is provided with a camera holder with stability, a camera can be used for obtaining video information of a scene, and then an image recognition technology and a data fusion technology are applied to automatically detect and analyze line defects. The detection of the target wire is the key that the robot with the motor can be used for successfully lapping the wire subsequently.
Currently, there are three main types of technical solutions for wire target detection: the wire target detection method based on the vision sensor, the wire target detection method based on the laser sensor and the wire target detection method based on the fusion of the vision sensor and the laser sensor.
The wire target detection method based on the vision sensor needs to acquire a wire target detection result by acquiring two-dimensional color image data of a wire and utilizing a target detection algorithm in the field of computer vision. The disadvantages of this approach are: the data acquired by the color camera is lack of three-dimensional position information of the wire target in the physical world.
The wire target detection method based on the laser sensor needs to obtain the position, end point and other related information of the wire target by analyzing the three-dimensional point cloud data acquired by the laser sensor. The disadvantages of this approach are: firstly, a relatively expensive laser radar is needed; secondly, due to the existence of noise and other obstacles, accurate electric wire target information is difficult to calculate, and the requirement on the surrounding environment is high; and thirdly, most of the wire detection algorithms based on the laser sensor use a global iteration method, so that the calculation amount is large, more calculation resources are needed, and the time is consumed.
The wire target detection method based on the fusion of the visual sensor and the laser sensor needs to accurately detect a wire target by using the visual sensor, and then determine other information such as a three-dimensional position, an end point and the like of the corresponding wire target by using the laser sensor. The method well overcomes the defects of using a laser sensor or a vision sensor independently, but needs to accurately mark the conversion relation between the two sensors, and meanwhile, the application of the two sensors can improve the cost of detecting the electric wire target.
Disclosure of Invention
The purpose of the invention is as follows: aiming at the defects of the prior art, the invention provides a wire target detection method based on Hough transform and feature space conversion. The method comprises the steps of converting depth data acquired by an area array laser sensor into a gray image, then carrying out data preprocessing on the gray image to acquire a binary image, carrying out Hough transformation to acquire a reference wire target in the binary image, then removing noise in the reference wire target in a feature space to acquire an effective reference wire target, and finally acquiring the position, end point information and confidence coefficient of the wire target in the corresponding depth data. Meanwhile, the sensor is an area array laser sensor, and a gray level camera in the sensor is originally used for area array laser calibration, so that the overall cost is low. The method of the invention can effectively solve the problems of the existing wire target detection method.
The technical scheme is as follows:
a wire target detection method comprises the following steps:
(1) acquiring original depth data of a scene where the electric wire is located through a laser sensor, and converting the original depth data into a gray image according to the distance of the depth;
(2) carrying out data preprocessing on the gray level image obtained in the step (1) to obtain a binary image;
(3) obtaining linear targets in all directions by using Hough transform on the binary image obtained in the step (2) and taking the linear targets as reference wire targets;
(4) converting the reference wire target obtained in the step (3) into an intercept-slope characteristic space, and denoising according to the slope and the intercept respectively to obtain an effective reference wire target;
(5) and (5) connecting the effective reference wire target obtained in the step (4) as a final wire target to obtain a wire target detection result.
The step (4) is specifically as follows:
(41) equally dividing the slope in the feature space into N groups, and classifying all the reference wire targets into corresponding groups according to the slope;
(42) determining N groups with the maximum number of corresponding reference wire targets as an effective group and the rest N-N groups as noise groups according to the number of the reference wire targets in each group, and removing the noise groups;
(43) equally dividing the intercepts in the feature space into M groups, and classifying the reference wire targets of the effective group in the step (42) into corresponding groups according to the intercept size;
(44) determining M groups with the maximum number of corresponding reference wire targets as an effective group and the rest M-M groups as noise groups according to the number of the reference wire targets in each group, and removing the noise groups;
(45) and (4) taking all the points in the effective groups reserved in the step (44) as effective point groups, respectively calculating the intercept mean value and the slope mean value of all the effective points to obtain a reference target of each effective point group, and determining K effective points which are closest to the reference target in each effective point group as effective reference wire targets of each effective point group.
And K is 3-5.
Further comprising a verification step:
calculating the confidence of the final wire target, wherein the confidence of the final wire target is the discrete degree of the effective reference wire target on the slope and the intercept respectively:
wherein, ai、ajRespectively representing the slopes of the ith and the jth significant point in the current significant point group, bi、bjRepresents the intercept of the ith and the jth effective points in the current effective point group.
The data preprocessing of the gray-scale image obtained in the step (1) specifically comprises the following steps: and performing smooth filtering, binary image extraction and noise removal on the gray level image respectively by adopting a bilateral filtering method, a Canny edge operator and parameter assistance of an area array laser sensor.
Has the advantages that:
1. compared with the wire target detection method based on the visual sensor, the method only needs depth data, and can directly acquire information in the physical world such as the position and the end point of the wire target;
2. compared with a wire target detection method based on a laser sensor, the method can be suitable for an area array laser sensor with lower cost, noise in depth data is effectively filtered by a digital image processing method, and meanwhile, calculation consumption is reduced;
3. compared with the wire target detection method based on the fusion of the visual sensor and the laser sensor, the method uses the area array laser sensor with lower cost, simultaneously reduces the step of fusion algorithm, and improves the calculation efficiency.
4. The method effectively utilizes the function of detecting the linear target of the wire in all directions by Hough transform, and reduces the possibility of missed detection; aiming at the reference wire targets in the effective class, the correlation among the reference wire targets is applied, the influence of noise is filtered, and the accuracy of the wire detection method is effectively improved. Finally, the invention provides each wire detection result and corresponding confidence coefficient, and further improves the accuracy of the wire detection result.
Drawings
FIG. 1 is a flow chart of the present invention;
fig. 2 is a schematic diagram of the display of a reference wire target in a binary image obtained by hough transform on a grayscale camera of an area-array laser;
FIG. 3 is an intercept-slope feature space diagram;
fig. 4 is a schematic diagram of obtaining a valid reference wire target.
Detailed Description
The invention is further elucidated with reference to the drawings and the embodiments.
FIG. 1 is a flow chart of the present invention. As shown in fig. 1, the wire target detection method of the present invention includes the steps of:
(1) acquiring original depth data of a scene through a laser sensor; in the invention, the laser sensor adopts an area array laser sensor.
(2) Converting the original depth data obtained in the step (1) into a gray image according to the depth distance; by adopting the method, noise can be effectively filtered, the target can be detected, and meanwhile, the calculated amount of two-dimensional data such as a gray image can be reduced compared with three-dimensional data.
(3) Carrying out data preprocessing on the gray level image obtained in the step (2) to obtain binary depth data; the step is used for extracting noise points in the gray level image and extracting foreground edge information in the gray level image;
the data preprocessing adopts a bilateral filtering method, a Canny edge operator and the assistance of relevant parameters of an area array laser sensor, and the data preprocessing respectively carries out smooth filtering on a gray level image, extracts a binary image and removes noise;
because the original depth data contains a large amount of noise, the accurate depth data part in the depth data needs to be used for assisting in noise filtering, a bilateral filtering method is utilized after the original depth data is converted into a gray image, and the method fully considers the relevant information of each pixel point in the horizontal and vertical directions and comprehensively gives a final filtering result;
in order to effectively distinguish the foreground from the background and improve the efficiency and accuracy of subsequent electric wire target detection, a Canny edge operator is used, the method can extract electric wires and other small amount of foreground information in the gray level image and provide a binary image, the efficiency of subsequent processing is effectively improved, and a large law algorithm is adopted to calculate relevant threshold parameters of the Canny edge operator, so that the reliability of the Canny edge operator is improved;
because the area array laser sensor has a measuring range, and meanwhile, the depth data of local noise points in the depth data is inaccurate and exceeds the measuring range, so that the detection accuracy is influenced, the related physical parameters of the area array laser sensor are used, pixels exceeding the measuring range of the area array sensor and corresponding gray level images are filtered, the accuracy of subsequent electric wire detection is improved, and the false detection rate is reduced.
(4) Obtaining linear targets in all directions by using Hough transform on the binary depth data obtained in the step (3) to be used as reference wire targets, and ensuring that the wire detection does not have the condition of missing detection; wherein, the reference wire target is a straight line target detected by hough transform, as shown in fig. 2;
(5) converting all the reference wire targets obtained in the step (4) into an intercept-slope feature space, wherein all the reference wire targets can be represented as one point in the intercept-slope feature space according to the intercept and the slope of the reference wire targets, and performing inter-group denoising on the points in all the feature spaces; as shown in fig. 3; the method specifically comprises the following steps:
(51) equally dividing the slope in the feature space into N groups, and classifying all the reference wire targets into corresponding groups according to the slope;
(52) determining N groups with the maximum number of corresponding reference wire targets as an effective group and the rest N-N groups as noise groups according to the number of the reference wire targets in each group, and removing the noise groups;
(53) equally dividing the intercept in the feature space into M groups, and classifying all the reference wire targets in the effective group obtained in the step (52) into corresponding groups according to the size of the intercept;
(54) determining M groups with the maximum number of corresponding reference wire targets as an effective group and the rest M-M groups as noise groups according to the number of the reference wire targets in each group, and removing the noise groups;
(55) all points in the valid groups reserved in the step (54) are used as valid point groups, namely points in the intersection parts of the slope and the intercept which are obtained after the slope and the intercept are respectively grouped and subjected to inter-group denoising in the feature space are used as valid point groups, the intercept mean value and the slope mean value of all valid points are respectively calculated to obtain the benchmark reference target of each valid point group, and K valid points which are closest to the benchmark reference target in each valid point group are determined to be used as valid reference wire targets of each valid point group, namely, a coarse-to-fine detection method is used for determining to obtain a fine target detection result;
in the invention, the distance among k effective points closest to the reference target in each effective point group is determined to be the Euclidean distance in a slope-intercept characteristic space, namely the sum of squared differences of different dimensions of each point in the slope-intercept characteristic space, and then the result of the power of one half is obtained.
In the invention, K is 3-5;
according to the invention, a cluster of points expressed as a straight line in the Euclidean space is expressed by two numerical values of intercept and slope, so that the calculation efficiency is improved, the data dimension is reduced, and the relation between all reference wire targets can be effectively found; and, according to the detection method from coarse to fine, noise among the reference electric wire targets is removed step by step.
(6) Connecting the effective reference wire targets in each effective point group as final wire targets based on the effective reference wire targets in each effective point group to obtain wire target detection results, and calculating to obtain the position and end point information of the wire target in the depth data;
(7) calculating the confidence of the final wire target;
the confidence of the final wire target is the discrete degree of the effective reference wire target on the slope and the intercept respectively:
wherein, ai、ajRespectively representing the slopes of the ith and the jth significant point in the current significant point group, bi、bjRepresents the intercept of the ith and the jth effective points in the current effective point group.
As can be seen from the above embodiments, the wire target detection method based on hough transform and feature space transformation of the present invention includes converting original depth data into a grayscale image and performing data preprocessing, removing noise existing in the data acquisition process, and distinguishing foreground and background information in the grayscale image; and after the conversion into the intercept-slope characteristic space, the data dimension is reduced, the interrelation among all the reference wire targets can be effectively found, the reference wire targets belonging to the same wire target are classified into the same category, and the effective reference wire targets are reserved for further calculation to obtain the final wire target. The classification method from coarse to fine ensures that the conditions of missing detection and false detection are effectively reduced in the process of detecting the electric wire target by the depth data. Finally, based on the correlation between the effective reference wire targets, the confidence coefficient of the detection result of each wire is calculated, the effectiveness of the detection result is further guaranteed, the situation that the actual wire target is too few or too complex is avoided, and the effectiveness, the accuracy and the reliability of the wire target detection method are improved.
Although the preferred embodiments of the present invention have been described in detail, the present invention is not limited to the details of the foregoing embodiments, and various equivalent changes (such as number, shape, position, etc.) may be made to the technical solution of the present invention within the technical spirit of the present invention, and these equivalent changes are all within the protection scope of the present invention.
Claims (5)
1. A wire target detection method is characterized by comprising the following steps:
(1) acquiring original depth data of a scene where the electric wire is located through a laser sensor, and converting the original depth data into a gray image according to the distance of the depth;
(2) carrying out data preprocessing on the gray level image obtained in the step (1) to obtain a binary image;
(3) obtaining linear targets in all directions by using Hough transform on the binary image obtained in the step (2) and taking the linear targets as reference wire targets;
(4) converting the reference wire target obtained in the step (3) into an intercept-slope characteristic space, and denoising according to the slope and the intercept respectively to obtain an effective reference wire target;
(5) and (5) connecting the effective reference wire target obtained in the step (4) as a final wire target to obtain a wire target detection result.
2. The wire target detection method according to claim 1, wherein the step (4) is as follows:
(41) equally dividing the slope in the feature space into N groups, and classifying all the reference wire targets into corresponding groups according to the slope;
(42) determining N groups with the maximum number of corresponding reference wire targets as an effective group and the rest N-N groups as noise groups according to the number of the reference wire targets in each group, and removing the noise groups;
(43) equally dividing the intercepts in the feature space into M groups, and classifying all the reference wire targets of the effective group in the step (42) into corresponding groups according to the intercept size;
(44) determining M groups with the maximum number of corresponding reference wire targets as an effective group and the rest M-M groups as noise groups according to the number of the reference wire targets in each group, and removing the noise groups;
(45) and (4) taking all the points in the effective groups reserved in the step (44) as effective point groups, respectively calculating the intercept mean value and the slope mean value of all the effective points to obtain a reference target of each effective point group, and determining K effective points which are closest to the reference target in each effective point group as effective reference wire targets of each effective point group.
3. The wire target detection method according to claim 2, wherein K is 3 to 5.
4. The wire target detection method of claim 1, further comprising a verification step of:
calculating the confidence of the final wire target, wherein the confidence of the final wire target is the discrete degree of the effective reference wire target on the slope and the intercept respectively:
wherein, ai、ajRespectively representing the slopes of the ith and the jth significant point in the current significant point group, bi、bjRepresents the intercept of the ith and the jth effective points in the current effective point group.
5. The wire target detection method according to claim 1, wherein the data preprocessing of the grayscale image obtained in step (1) is specifically: and performing smooth filtering, binary image extraction and noise removal on the gray level image respectively by adopting a bilateral filtering method, a Canny edge operator and parameter assistance of an area array laser sensor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110158342.6A CN113160118A (en) | 2021-02-05 | 2021-02-05 | Wire target detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110158342.6A CN113160118A (en) | 2021-02-05 | 2021-02-05 | Wire target detection method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113160118A true CN113160118A (en) | 2021-07-23 |
Family
ID=76882928
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110158342.6A Pending CN113160118A (en) | 2021-02-05 | 2021-02-05 | Wire target detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113160118A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2741233A2 (en) * | 2012-12-04 | 2014-06-11 | Ricoh Company, Ltd. | Method and apparatus for detecting road |
| 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 |
| US20190206043A1 (en) * | 2017-12-29 | 2019-07-04 | Huizhou China Star Optoelectronics Technology Co., Ltd. | Method and system for detection of in-panel mura based on hough transform and gaussian fitting |
| CN110134148A (en) * | 2019-05-24 | 2019-08-16 | 中国南方电网有限责任公司超高压输电公司检修试验中心 | A kind of transmission line of electricity helicopter make an inspection tour in tracking along transmission line of electricity |
-
2021
- 2021-02-05 CN CN202110158342.6A patent/CN113160118A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2741233A2 (en) * | 2012-12-04 | 2014-06-11 | Ricoh Company, Ltd. | Method and apparatus for detecting road |
| US20190206043A1 (en) * | 2017-12-29 | 2019-07-04 | Huizhou China Star Optoelectronics Technology Co., Ltd. | Method and system for detection of in-panel mura based on hough transform and gaussian fitting |
| 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 |
| CN110134148A (en) * | 2019-05-24 | 2019-08-16 | 中国南方电网有限责任公司超高压输电公司检修试验中心 | A kind of transmission line of electricity helicopter make an inspection tour in tracking along transmission line of electricity |
Non-Patent Citations (1)
| Title |
|---|
| 方苏;李立学;郑益慧;王昕;于建友;杨景波;: "基于激光测距成像和图像处理的输电线路防护技术", 电气自动化, no. 03, pages 6 - 8 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110070570B (en) | Obstacle detection system and method based on depth information | |
| CN110647850A (en) | Automatic lane deviation measuring method based on inverse perspective principle | |
| CN110033431B (en) | Non-contact detection device and detection method for detecting corrosion area on surface of steel bridge | |
| CN111369516B (en) | Transformer bushing heating defect detection method based on infrared image recognition | |
| CN106960207A (en) | A kind of car steering position gauge field multipointer instrument automatic recognition system and method based on template matches | |
| CN108921813B (en) | Unmanned aerial vehicle detection bridge structure crack identification method based on machine vision | |
| CN110910350B (en) | Nut loosening detection method for wind power tower cylinder | |
| KR20150112656A (en) | Method to calibrate camera and apparatus therefor | |
| CN115359021A (en) | Target positioning detection method based on laser radar and camera information fusion | |
| CN110222661B (en) | Feature extraction method for moving target identification and tracking | |
| CN114331986A (en) | Dam crack identification and measurement method based on unmanned aerial vehicle vision | |
| CN109063669B (en) | Bridge area ship navigation situation analysis method and device based on image recognition | |
| CN115731257A (en) | Leaf form information extraction method based on image | |
| CN117197700B (en) | Intelligent unmanned inspection contact net defect identification system | |
| CN104966283A (en) | Imaging layered registering method | |
| CN113723314A (en) | Sugarcane stem node identification method based on YOLOv3 algorithm | |
| Raikar et al. | Automatic building detection from satellite images using internal gray variance and digital surface model | |
| CN114758139B (en) | Method for detecting accumulated water in foundation pit | |
| CN113066041A (en) | Pavement crack detection method based on stack sparse self-coding deep learning | |
| CN114283126B (en) | Method for detecting deviation of monitoring equipment of power transmission line | |
| CN112734745B (en) | Unmanned aerial vehicle thermal infrared image heating pipeline leakage detection method fusing GIS data | |
| CN113160118A (en) | Wire target detection method | |
| Sun et al. | A vision system based on TOF 3D imaging technology applied to robotic citrus harvesting | |
| CN113450335B (en) | Road edge detection method, road edge detection device and road surface construction vehicle | |
| CN114677428A (en) | Power transmission line icing thickness detection method based on unmanned aerial vehicle image processing |
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 |