CN111292376B - Visual target tracking method of bionic retina - Google Patents
Visual target tracking method of bionic retina Download PDFInfo
- Publication number
- CN111292376B CN111292376B CN202010090894.3A CN202010090894A CN111292376B CN 111292376 B CN111292376 B CN 111292376B CN 202010090894 A CN202010090894 A CN 202010090894A CN 111292376 B CN111292376 B CN 111292376B
- Authority
- CN
- China
- Prior art keywords
- target
- information
- image
- scanning
- imaging
- 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.)
- Active
Links
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
Abstract
The invention discloses a visual target tracking method of a bionic retina, belonging to the field of photoelectric target tracking. The invention mainly uses the variable resolution characteristic of bionic retina imaging to map the view field into the logarithmic polar coordinates for target tracking. Firstly, utilizing the logarithmic polar coordinate mapping imaging of the bionic retina, improving a target tracking algorithm through the gaze action of a central recess, the information compression action of an edge and real-time parameters, compressing redundant data, enhancing the detail information of an interested target and improving the information acquisition efficiency in a video field range. Meanwhile, the singular point and the imaging threshold value are used for judging, the target information in the scene is accurately grasped, and the reliability of target screening is improved. Finally, real-time target tracking is realized through comparing an eccentric judgment algorithm with a motion control module with a higher speed. Compared with the traditional method, the method realizes real-time tracking of the target based on the bionic retina vision mechanism under the condition of large visual field and high resolution, and has higher efficiency and higher precision.
Description
Technical Field
The invention belongs to the field of photoelectric target tracking, and particularly relates to a visual target tracking method of a bionic retina.
Background
Target recognition and tracking is a common technology, the core of which is to locate and complete tracking in real time for a specific or interested object, and the technology is widely applied to various occasions at present, for example: in the field of robots, visual perception is a prerequisite for many applications of robots (navigation, obstacle avoidance, etc.), and a requirement for large field of view, high resolution and real-time is generally required for visual perception, which puts a severe requirement on traditional visual perception. On the other hand, when the robot performs object recognition, the relative size of the object under the condition of large field of view is small, so that the background except the object becomes redundant data, and the real-time performance is reduced, and therefore, the accurate perception and object tracking of the small object become the difficulty in the field of robots. In the field of industrial detection, workpiece flaw detection is a typical and indispensable link in intelligent industrial manufacturing, for nonstandard products, the sizes of workpieces are different, the materials are often different, a traditional fixed resolution sensing system can only fully image scenes in a field of view, and although the workpiece flaw detection can be completed to a certain extent by matching with a corresponding algorithm, the real-time discovery and acquisition of specific flaw and damaged parts are difficult, namely the intelligent degree is still to be improved, so that the detection performance of the local flaw, defect and damaged parts in the field of industrial measurement and detection is also required to be high in resolution and real-time. In the biomedical field, a specific cell is often required to be observed in real time, but because the motion range of a living cell is wider and the living cell has three-dimensional characteristics, the living cell is required to be obtained in two dimensions and three dimensions, and under the same time condition, three-dimensional data is larger than the information quantity of two-dimensional data, which means that real-time performance is more difficult to meet in a large-view and high-resolution occasion, so that the biological field has higher requirements for high-performance target tracking.
At present, typical target tracking is to detect a single object moving in a video sequence or having a certain characteristic through an algorithm, acquire parameters such as coordinates, speed and the like of the single object, and then track the single object.
The above shows that: although the traditional target tracking method can realize tracking to a certain extent, the problems of poor real-time performance and poor reliability exist.
Disclosure of Invention
The invention discloses a visual target tracking method of a bionic retina, which aims to solve the technical problems that: under the condition of large visual field and high resolution, real-time tracking of the target is realized based on a bionic retina vision mechanism, and the method has the advantages of high efficiency and high precision.
The invention aims at realizing the following technical scheme:
the invention discloses a visual target tracking method of a bionic retina, which mainly utilizes the variable resolution characteristic of bionic retina imaging to map a field of view into a logarithmic polar coordinate for target tracking. Firstly, utilizing the logarithmic polar coordinate mapping imaging of the bionic retina, improving a target tracking algorithm through the gaze action of a central recess, the information compression action of an edge and real-time parameters, compressing redundant data, enhancing the detail information of an interested target and improving the information acquisition efficiency in a video field range. Meanwhile, the singular point and the imaging threshold value are used for judging, the target information in the scene is accurately grasped, and the reliability of target screening is improved. Finally, real-time target tracking is realized through comparing an eccentric judgment algorithm with a motion control module with a higher speed. Compared with the traditional method, the method realizes real-time tracking of the target based on the bionic retina vision mechanism under the condition of large visual field and high resolution, and has higher efficiency and higher precision.
The invention discloses a visual target tracking method of a bionic retina, which comprises the following steps:
step one: according to the required approximate position and size of the detected and tracked object, the initial parameters of the bionic retina scanning are adjusted. Ensuring that the object of interest is present within the scan range.
Step two: an initial scan is started. Initial bionic retina scanning is carried out by presetting an initial threshold value, so that an initial image with a larger visual field range and a wider scanning range but a lower data volume is obtained, and target information is quickly searched in a large visual field range.
The implementation method of the second step is as follows: the initial number of scanning loops M, the number of blocks per loop N and the inter-loop growth rate q are preset according to empirical parameters. According to the following bionic retina scanning model
An initial image with a larger field of view and a wider scanning range but a lower data volume can be obtained for quickly searching for target information in a large field of view.
Wherein: zeta type toy i And theta j Alpha is a parameter in corresponding logarithmic polar coordinates j Is the included angle between the connecting line from the center of the jth pixel in a certain ring to the origin and the X axis. The radius of the blind hole is r 0 The radius of the circle where the ith pixel is positioned is r i The diameter of the pixel in the ith ring is D i The maximum radius of the sampling model is r max . In order to obtain higher focusing accuracy, the sampling loss of pixels of an image needs to be reduced to the minimum, namely, the pixels in adjacent rings are tangent, each pixel in the same ring is tangent, and the included angle between two adjacent pixels in the same ring is 2 pi/N.
Because of the redundant compression effect of the logarithmic polar coordinates, the data compression effect is obvious at the edge position with lower possibility of containing target information; at the central position with higher possibility of containing target information, the staring action of the central recess can acquire clearer target detail information, so that low data operand imaging under the condition of large field of view can be achieved, and the imaging efficiency is greatly improved.
Step three: and (3) selecting a target judgment mode for the image obtained by the second scanning, ensuring that the selected target judgment mode can accurately select target information, and checking whether singular data exists in the target information so as to meet the specification requirements of subsequent target processing.
The implementation method of the third step is as follows: classifying according to a target judging mode, namely a manual mode and an automatic mode, wherein the manual mode can directly select targets of scanned images, and then directly judging in a fourth step; the automatic method is to perform image recognition through various ways such as depth information, shape information and the like, and acquire required target information from an image. And after obtaining the image information, carrying out singular data judgment on the acquired image. By the following formula
Where m is a threshold designed according to an actual scene, f (x, y) is a depth information value of the point, Δ is a representative amount of a depth information difference, and H (x, y) is a representative amount of a difference between the depth information of each pixel point and the surrounding depth information. When the depth information of 24 points around the point has larger difference with the point, namely the H (x, y) value is larger, the point singular point can be judged. If obvious abnormal singular data exists in the scanned image, a fifth step is needed; if no singular data exists, the image information obtained by collecting the interested target under the parameter condition is proved to be complete and reliable, and the fourth operation can be performed.
Step four: and (3) judging the target information of the initial image in the step two, and comparing the initial image with a judging threshold value to ensure that the image contains sufficient target detail information and accords with the subsequent processing specification.
The implementation method of the fourth step is as follows: and judging target information of the image. Because of the compression effect of bionic retina imaging on the image edge area, the interested target with serious eccentricity can be in the scanning range of the field of view, but the selection of the number of scanning rings M, the number of blocks N of each ring or the inter-ring growth rate q is poor, the initial image imaging quality is poor in the second step, and effective detail information is difficult to obtain. Therefore, a judging threshold value is added, and when the imaging pixel number of the interested target in the field of view is larger than the lowest threshold value, the imaging is proved to accord with the specification requirement of the subsequent target, and the step six can be performed; otherwise, the adjustment of the fifth step is needed.
Step five: and (3) under the condition that the corresponding specifications in the third and fourth steps are not met, rescanning and imaging are carried out by adjusting imaging parameters so as to obtain images with better quality, wherein the images meet the corresponding specifications in the third and fourth steps.
The fifth implementation method comprises the following steps: judging the target identified in the step three, and when the specification requirement of the step four is not met; or when the singular data exists in the automatic target judging method, the parameters are reset for scanning. According to the target characteristics, when the target occupies less space in the radial direction (i.e., ρ coordinate axis), the number of scanning loops M and the inter-loop growth rate q are increased appropriately; when the space occupied by the target in the circumferential direction (namely, theta coordinate axis) is small, the imaging effect of the target in the bionic retina logarithmic polar coordinate system is better by properly increasing the number of blocks N of each ring. And when the imaging pixel number of the target can reach the preset number, the positioning accuracy of the mass center of the target can be correspondingly improved. And after adjusting the corresponding imaging parameters, returning to the second step for processing. The radial direction is the rho coordinate axis direction, and the circumferential direction is the theta coordinate axis direction.
Step six: and (3) judging the target position information of the image conforming to the specification in the step (IV), namely judging whether the target is eccentric or not by comparing and judging the rho coordinates, and executing the step (seven) or the step (eight) according to the judgment result.
The sixth implementation method comprises the following steps: and when the identified target is judged in the third step and the judgment result reaches the fourth threshold value standard, the image is a better imaging image containing the target position information. And on the basis, the eccentricity judgment of the target is performed. Because of the excellent property of the logarithmic polar coordinates, after the centroid of the target is calculated, the ρ coordinates are only required to be compared and judged. Compared with the Cartesian coordinate system, the X coordinate axis and the Y coordinate axis are required to be respectively judged, and the judgment process can reduce the calculated amount by one time. If the target is eccentric, performing the operation of the step seven; otherwise, step eight is performed.
Step seven: if the target is eccentric in the acquired image, the tracking operation of the target is needed, the target position information data obtained in the step six is output, the platform of the scanning device is moved to the corresponding position by controlling the motion module, and the step one is returned to scan.
The seventh implementation method comprises the following steps: if the eccentric condition exists in the acquired image, the tracking operation of the target is needed, and the (rho, theta) data obtained by calculation in the step six is output, and the scanning is performed by controlling the motion module to move the platform of the scanning device to the corresponding position in combination with the setting of parameters M, N and q during scanning.
Step eight: if the target is not eccentric in the acquired images, outputting the images of the target, and ending the scanning imaging process, namely realizing high-precision and high-efficiency real-time tracking of the target based on a bionic retina vision mechanism under the condition of large visual field and high resolution.
The beneficial effects are that:
1. according to the visual target tracking method of the bionic retina, due to the redundant compression effect of logarithmic polar coordinates, the data compression effect is obvious at the edge position with low possibility of containing target information; in the central position with higher possibility of containing target information, the staring action of the central recess can acquire clearer target detail information, and under the condition of large view field, low data operand imaging under the condition of large view field can be achieved through logarithmic polar coordinate mapping, and imaging efficiency is greatly improved.
2. According to the visual target tracking method of the bionic retina, through the logarithmic polar coordinate structure property, the interested target is automatically judged in the near-sighted central area, the rotation and scale invariance property is achieved, and the two-dimensional gesture limitation on the target is small.
3. According to the visual target tracking method of the bionic retina, through the logarithmic polar coordinate structure property, the eccentric position of the bionic retina is judged by only comparing and judging rho coordinates. Compared with the Cartesian coordinate system, the X coordinate axis and the Y coordinate axis are required to be respectively judged, and the judgment process can reduce the calculated amount by one time.
Drawings
Fig. 1 is a log polar mapping schematic diagram in which: FIG. 1 (a) is a Cartesian coordinate plane and FIG. 1 (b) is a logarithmic polar coordinate plane;
fig. 2 is a workflow diagram of a method for visual target tracking of a simulated retina according to the present disclosure.
FIG. 3 is a schematic diagram of a tracking method according to an embodiment of the present invention, wherein FIG. 3 a) is an eccentric five-pointed star target original state diagram, 3 b) is a five-pointed star target positioning diagram after log-polar coordinate mapping, and 3 c) is a tracking result diagram
Detailed Description
For a better description of the objects and advantages of the present invention, the following description will be given with reference to the accompanying drawings and examples.
Example 1:
as shown in fig. 2, the embodiment discloses a visual target tracking method for a bionic retina, which specifically includes the following implementation steps:
step one: according to the required approximate position and size of the detected and tracked object, the initial parameters of the bionic retina scanning are adjusted. Ensuring that the object of interest is present within the scan range. The scene is shown in fig. 3 a), the whole is 600×600 pixels, and the tracking target is the upper left five-pointed star target.
Step two: an initial scan is started. Initial bionic retina scanning is carried out by presetting an initial threshold value, so that an initial image with a larger visual field range and a wider scanning range but a lower data volume is obtained, and target information is quickly searched in a large visual field range.
According to the empirical parameters, the initial scanning loop number M is preset to 400, the block number N of each loop is 400, and the inter-loop growth rate q is 1.0158. According to the following bionic retina scanning model
An initial image with a larger field of view and a wider scanning range but a lower data volume can be obtained for quickly searching for target information in a large field of view. Forming a log polar image as shown in fig. 3 b).
Step three: and (3) selecting a target judgment mode for the image obtained by the second scanning, ensuring that the selected target judgment mode can accurately select target information, and checking whether singular data exists in the target information so as to meet the specification requirements of subsequent target processing.
This example uses an automatic identification method. The automatic method is to perform image recognition through various ways such as depth information, shape information and the like, and acquire required target information from an image. And after obtaining the image information, carrying out singular data judgment on the acquired image.
By the following formula
Where m is a threshold designed according to an actual scene, f (x, y) is a depth information value of the point, Δ is a representative amount of a depth information difference, and H (x, y) is a representative amount of a difference between the depth information of each pixel point and the surrounding depth information. When the depth information of 24 points around the point has larger difference with the point, namely the H (x, y) value is larger, the point singular point can be judged. The present example selects a threshold m of 50 and the criteria for h (x, y) of 20. According to the calculation of each pixel in the figure, H (x, y) of each point of the image is below 17, no singular point exists, and the information judgment threshold comparison can be continued.
Step four: and (3) judging the target information of the initial image in the step two, and comparing the initial image with a judging threshold value to ensure that the image contains sufficient target detail information and accords with the subsequent processing specification.
And judging target information of the image. Because of the compression effect of bionic retina imaging on the image edge area, the interested target with serious eccentricity can be in the scanning range of the field of view, but the selection of the number of scanning rings M, the number of blocks N of each ring or the inter-ring growth rate q is poor, the initial image imaging quality is poor in the second step, and effective detail information is difficult to obtain. Therefore, adding a decision threshold, wherein the number of rings occupied by the target five-pointed star in the example is 337 to 389, and the number of blocks occupied is 146 to 199, so that the distribution is reasonable; there are 1110 pixels in total, which may contain sufficient information. Therefore, the scanning parameters are reasonably selected, the scanning imaging result meets the requirement, and the position judgment can be directly carried out.
Step five: and (3) judging the target position information of the image conforming to the specification in the step (IV), and judging whether the target is eccentric or not by only comparing and judging the rho coordinates.
The determination is made that the image contains a preferred imaged image of the target position information, on the basis of which the decentration determination of the target is made. Because of the excellent property of the logarithmic polar coordinates, after the centroid of the target is calculated, the ρ coordinates are only required to be compared and judged. The finding of the target centroid is shown in fig. 3 b). The ρ coordinate is not 0, and the object has eccentricity. Further calculation is performed, and the eccentric coordinates thereof are found to be (349,171).
Step six: and if the target is eccentric, the tracking operation of the target is required, the target position information data obtained in the fifth step is output, the platform of the scanning device is moved to the corresponding position by controlling the motion module, and the first step is returned to scan.
If the target is eccentric in the acquired image, the tracking operation of the target is needed, and therefore the data (349,171) obtained by calculation in the step six is output, and the step one is returned by controlling the motion module to move the platform of the scanning device to the corresponding position in combination with the parameter setting m=n=400 and q= 1.0158 during scanning.
Step eight: and (3) scanning the tracking effect image again in the second step to obtain the tracking effect image shown in fig. 3 c). And (3) through the calculation of the third to fifth steps, the target has no eccentric condition, namely the eccentric coordinates are (0, 0), the image of the result can be output, and the scanning imaging process is finished, namely the high-precision and high-efficiency real-time tracking of the target is realized based on the bionic retina vision mechanism under the condition of large visual field and high resolution.
While the foregoing is directed to embodiments of the present invention, other and further embodiments of the invention may be devised without departing from the basic scope thereof, and the scope thereof is determined by the claims that follow.
Claims (6)
1. A visual target tracking method of a bionic retina is characterized by comprising the following steps of: comprises the following steps of the method,
step one: according to the approximate position and the size of the object to be detected and tracked, initial parameters of bionic retina scanning are adjusted; ensuring that the object of interest appears within the scanning range;
step two: starting initial scanning; initial bionic retina scanning is carried out by presetting an initial threshold value, so that an initial image with a larger visual field range and a wider scanning range but lower data volume is obtained, and target information is quickly searched in a large visual field range;
step three: selecting a target judgment mode for the image obtained by the second scanning, ensuring that the selected target judgment mode can accurately select target information, and checking whether singular data exists in the target information so as to meet the specification requirements of subsequent target processing;
step four: judging target information of the initial image in the second step, and comparing the initial image with a judging threshold value to ensure that the image contains sufficient target detail information and accords with the subsequent processing specification;
step five: for the condition that the corresponding specifications in the third and fourth steps are not met, re-scanning imaging is carried out by adjusting imaging parameters so as to obtain images with better quality, wherein the images meet the corresponding specifications in the third and fourth steps;
judging the target identified in the step three, and when the standard requirement of the step four is not met; or when the singular data exists in the automatic target judging method, resetting parameters for scanning; according to the target characteristics, when the target occupies less space in the radial direction, the number of scanning loops M and the inter-loop growth rate q are increased appropriately; when the space occupied by the target in the circumferential direction is small, the imaging effect of the target in the bionic retina logarithmic polar coordinate system is better by properly increasing the number N of blocks of each ring; when the imaging pixel number of the target reaches a preset number, the positioning accuracy of the mass center of the target is correspondingly improved; after adjusting the corresponding imaging parameters, returning to the second step for processing; the radial direction is rho coordinate axis direction, and the circumferential direction is theta coordinate axis direction;
step six: the image conforming to the specification in the fourth step is subjected to target position information judgment, whether the target is eccentric or not can be judged by only comparing and judging the rho coordinates, and the seventh step or the eighth step is executed according to the judgment result;
step seven: if the acquired image has eccentric condition, the tracking operation of the target is needed, the target position information data obtained by calculation in the step six is output, the platform of the scanning device is moved to the corresponding position by controlling the motion module, and the step one is returned to scan;
step eight: if the target is not eccentric in the acquired images, outputting the images of the target, and ending the scanning imaging process, namely realizing high-precision and high-efficiency real-time tracking of the target based on a bionic retina vision mechanism under the condition of large visual field and high resolution.
2. A method for tracking a visual target of a simulated retina as claimed in claim 1, wherein: the second implementation method is that the initial scanning ring number M, the block number N of each ring and the inter-ring growth rate q are preset according to experience parameters; according to the following bionic retina scanning model
The method can obtain an initial image with a larger visual field range, a wider scanning range and a lower data volume, and is used for rapidly searching target information in a large visual field range;
wherein: zeta type toy i And theta j Alpha is a parameter in corresponding logarithmic polar coordinates j An included angle between a connecting line from the center of a jth pixel in a certain ring to an origin and an X axis; the radius of the blind hole is r 0 The radius of the circle where the ith pixel is positioned is r i The diameter of the pixel in the ith ring is D i The maximum radius of the sampling model is r max The method comprises the steps of carrying out a first treatment on the surface of the In order to obtain higher focusing accuracy, the sampling loss of the pixels of the image needs to be reduced to the minimum, namely, the pixels in adjacent rings are tangent, each pixel in the same ring is tangent, and the included angle between two adjacent pixels in the same ring is 2 pi/N;
because of the redundant compression effect of the logarithmic polar coordinates, the data compression effect is obvious at the edge position with lower possibility of containing target information; and in the central position with higher possibility of containing the target information, the central concave staring is used for acquiring clearer target detail information, so that low data operand imaging under the condition of large field of view can be achieved, and the imaging efficiency is greatly improved.
3. A method for tracking a visual target of a simulated retina as claimed in claim 2, wherein: the third implementation method is that the target judgment is classified into a manual mode and an automatic mode, wherein the manual mode can directly select the target of the scanned image, and then the judgment of the fourth step is directly carried out; the automatic mode is that image recognition is carried out through various ways such as depth information, shape information and the like, and required target information is obtained from an image; after obtaining the image information, carrying out singular data judgment on the acquired image; by the following formula
Wherein m is a threshold designed according to an actual scene, f (x, y) is a depth information value of the point, delta is a representative amount of depth information difference, and H (x, y) is a representative amount of difference between the depth information of each pixel point and peripheral depth information; when the depth information of 24 points around the point has larger difference with the point, namely the H (x, y) value is larger, namely the point singular point is judged; if obvious abnormal singular data exists in the scanned image, a fifth step is needed; if no singular data exists, the image information obtained by collecting the interested target under the parameter condition is proved to be complete and reliable, and the fourth operation can be performed.
4. A method for tracking a visual target of a simulated retina as claimed in claim 3, wherein: the fourth implementation method is that target information judgment is carried out on the image; because of the compression effect of bionic retina imaging on the image edge area, the interested target with serious eccentricity can be in the scanning range of the field of view, but the selection of the number N of blocks of each ring or the inter-ring growth rate q is poor due to the number M of scanning rings, the initial image imaging quality is poor in the second step, and effective detail information is difficult to obtain; therefore, a judging threshold value is added, and when the imaging pixel number of the interested target in the field of view is larger than the lowest threshold value, the imaging is proved to accord with the specification requirement of the subsequent target, and the step six can be performed; otherwise, the adjustment of the fifth step is needed.
5. The method for tracking a visual target of a bionic retina according to claim 4, wherein: the specific implementation method of the sixth step is that when the recognized target is judged in the third step and the judgment result reaches the fourth threshold value standard, the image is a better imaging image containing the target position information; performing eccentric judgment of the target on the basis; because of the excellent property of the logarithmic polar coordinates, after the mass center of the target is calculated, the rho coordinates are only required to be compared and judged, and if the target is eccentric, the operation of the step seven is carried out; otherwise, step eight is performed.
6. The method for tracking a visual target of a bionic retina according to claim 5, wherein: and step seven, if the eccentric condition exists in the acquired image, the tracking operation of the target is needed, the (rho, theta) data calculated and obtained in step six are output, the parameters M, N and q during scanning are combined, the platform of the scanning device is moved to the corresponding position by controlling the motion module, and the step one is returned to carry out scanning.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010090894.3A CN111292376B (en) | 2020-02-13 | 2020-02-13 | Visual target tracking method of bionic retina |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010090894.3A CN111292376B (en) | 2020-02-13 | 2020-02-13 | Visual target tracking method of bionic retina |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111292376A CN111292376A (en) | 2020-06-16 |
| CN111292376B true CN111292376B (en) | 2023-06-30 |
Family
ID=71028506
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010090894.3A Active CN111292376B (en) | 2020-02-13 | 2020-02-13 | Visual target tracking method of bionic retina |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111292376B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113109833A (en) * | 2021-04-02 | 2021-07-13 | 北京理明智能科技有限公司 | Bionic three-dimensional imaging system and method based on fusion of visible light and laser radar |
| CN113518214B (en) * | 2021-05-25 | 2022-03-15 | 上海哔哩哔哩科技有限公司 | Panoramic video data processing method and device |
| CN114993262B (en) * | 2022-04-20 | 2023-04-18 | 北京航空航天大学 | Sea surface unmanned ship target identification method imitating reconstruction of bird prey receptive field area |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101349765A (en) * | 2008-08-29 | 2009-01-21 | 北京理工大学 | Photodetection method based on axial vibration retina |
| CN109974853A (en) * | 2018-12-28 | 2019-07-05 | 北京信息科技大学 | Based on the multispectral compound detection of bionical sensation target and tracking |
-
2020
- 2020-02-13 CN CN202010090894.3A patent/CN111292376B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101349765A (en) * | 2008-08-29 | 2009-01-21 | 北京理工大学 | Photodetection method based on axial vibration retina |
| CN109974853A (en) * | 2018-12-28 | 2019-07-05 | 北京信息科技大学 | Based on the multispectral compound detection of bionical sensation target and tracking |
Non-Patent Citations (2)
| Title |
|---|
| 基于人类视觉特性的机器视觉系统;毛晓波;陈铁军;;仪器仪表学报(第04期);114-119 * |
| 基于对数极坐标映射的目标跟踪方法;陈朝阳,丁明跃,周成平;电子学报(第09期);1173-1175 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111292376A (en) | 2020-06-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111292376B (en) | Visual target tracking method of bionic retina | |
| CN110555889B (en) | CALTag and point cloud information-based depth camera hand-eye calibration method | |
| CN109308693B (en) | Single-binocular vision system for target detection and pose measurement constructed by one PTZ camera | |
| CN105729468B (en) | A kind of robotic workstation based on the enhancing of more depth cameras | |
| CN107471218B (en) | Binocular vision-based hand-eye coordination method for double-arm robot | |
| CN107993258B (en) | Image registration method and device | |
| CN110211180A (en) | A kind of autonomous grasping means of mechanical arm based on deep learning | |
| CN104331894A (en) | Robot unstacking method based on binocular stereoscopic vision | |
| CN105217324A (en) | A kind of novel de-stacking method and system | |
| CN105666274B (en) | A kind of service plate edging method of view-based access control model control | |
| CN111402330A (en) | Laser line key point extraction method based on plane target | |
| CN108074265A (en) | A kind of tennis alignment system, the method and device of view-based access control model identification | |
| CN115171097A (en) | Processing control method and system based on three-dimensional point cloud and related equipment | |
| CN111583342A (en) | Target rapid positioning method and device based on binocular vision | |
| CN113034526B (en) | Grabbing method, grabbing device and robot | |
| CN109313708B (en) | Image matching method and vision system | |
| CN111684462A (en) | Image matching method and visual system | |
| AU2021105660A4 (en) | Visual target tracking method based on bionic retina | |
| CN116486287A (en) | Target detection method and system based on environment self-adaptive robot vision system | |
| CN116228783A (en) | Workpiece point cloud segmentation method under complex background and application thereof | |
| CN109211222B (en) | High-precision positioning system and method based on machine vision | |
| CN115034577A (en) | Electromechanical product neglected loading detection method based on virtual-real edge matching | |
| CN113932712A (en) | Melon and fruit vegetable size measuring method based on depth camera and key points | |
| CN114549420A (en) | Workpiece identification and positioning method based on template matching | |
| CN107145906B (en) | Mobile robot indoor rapid homing method based on panoramic visual imaging system |
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 |