CN112351181A - Intelligent camera based on CMOS chip and ZYNQ system - Google Patents
Intelligent camera based on CMOS chip and ZYNQ system Download PDFInfo
- Publication number
- CN112351181A CN112351181A CN202011251121.5A CN202011251121A CN112351181A CN 112351181 A CN112351181 A CN 112351181A CN 202011251121 A CN202011251121 A CN 202011251121A CN 112351181 A CN112351181 A CN 112351181A
- Authority
- CN
- China
- Prior art keywords
- image
- video
- zynq
- hardware unit
- pixel
- 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
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
- H04N23/50—Constructional details
- H04N23/54—Mounting of pick-up tubes, electronic image sensors, deviation or focusing coils
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/12—Edge-based segmentation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
- H04N23/80—Camera processing pipelines; Components thereof
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Signal Processing (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention provides an intelligent camera based on a CMOS chip and a ZYNQ system, which comprises a ZYNQ platform and a CMOS sensor, wherein the ZYNQ platform is connected with the CMOS sensor through a ZYNQ interface; the ZYNQ platform comprises an FPGA hardware unit and an ARM hardware unit; the FPGA hardware unit comprises a video acquisition module and a video preprocessing module; the video acquisition module acquires video signal information through the CMOS sensor and sends the video signal information to the video preprocessing module, and the video preprocessing module sends the processed video signal to the ARM hardware unit; the ARM hardware unit runs a linux operating system, and an application layer video image processing module is installed in the linux operating system. The intelligent camera hardware system has the advantages that the intelligent camera hardware system provided by the invention designs the image processing application program on the application layer to realize the video data flow control of FPGA hardware processing and ARM cooperative processing, completes the image processing of a large number of involved image processing algorithms and realizes the real-time acquisition and processing of high-definition video data streams.
Description
Technical Field
The invention relates to the field of machine vision, in particular to an intelligent camera based on a CMOS chip and a ZYNQ system.
Background
The machine vision detection technology plays a very important role in the field of industrial application, particularly the field of industrial robots, and as an important input channel for sensing external environment information of the industrial robot, the machine vision detection technology plays a very important role in understanding surrounding scenes and assisting in completing specific tasks of the industrial robot. At present, the application of the visual recognition technology in the field of robots mainly comprises environment understanding, self-learning object recognition and intelligent interaction, navigation, obstacle avoidance and the like.
In the field of machine vision, particularly in the field of mechanical measurement based on industrial robots, the existing machine vision detection technology has the defects that a measurement system is relatively independent, the construction is complex, the system is huge, and the moving and the construction are inconvenient; the vision measuring system is fixed outside the industrial robot (eye-to-hand) without moving with the arm, and the problems of low intelligent degree, poor applicability and the like such as large system error and the like are solved.
Disclosure of Invention
The invention aims to solve the technical problems that the existing machine vision detection technology has relatively independent measurement systems, complex construction, huge system and inconvenience in moving and building in the field of machine vision, especially in the field of mechanical measurement based on industrial robots; the vision measuring system is fixed outside the industrial robot (eye-to-hand) without moving with the arm, and has low intelligent degree and poor applicability such as large system error.
The invention provides an intelligent camera based on a CMOS chip and a ZYNQ system, which comprises,
ZYNQ platform, CMOS sensor;
the ZYNQ platform comprises an FPGA hardware unit and an ARM hardware unit;
the FPGA hardware unit comprises a video acquisition module and a video preprocessing module;
the video acquisition module acquires video signal information through the CMOS sensor and sends the video signal information to the video preprocessing module, and the video preprocessing module sends the processed video signal to the ARM hardware unit;
the ARM hardware unit runs a linux operating system, and an application layer video image processing module is installed in the linux operating system.
Further, the application layer video image processing module executes a template matching target based search algorithm,
the template matching based object finding algorithm comprises the following steps,
an image preprocessing step:
wherein f (x, y) is the preprocessed image; θ is a local neighborhood of the current pixel (m, n). One-sided leave-on function u exists with an inverse function u-1A (i, j) is a weighting coefficient, g (i, j) is an input image;
template matching:
s (m, n) is a template image, M, N is the dimension of the template image, f (m, n) is a sub-image in the f (x, y) image with the same size as the template image, and D (x, y) is the measure of matching error;
a geometric transformation step:
mapping the template image to the position of the processed image through geometric transformation,
t is a vector function, (x, y) is the pixel coordinates of the template image, and (x ', y') is the new coordinates of the transformed pixels in the processed image.
Further, the application layer video image processing module executes an edge extraction segmentation algorithm,
the edge extraction segmentation algorithm comprises the following steps,
an edge extraction step:
the image is first convolved with a gaussian function of scale sigma,
secondly, for each pixel in the image, the normal n of the local edge is estimated,
the position of the edge is found again and,
the edge strength is calculated again and the edge strength is calculated,
finally, hysteresis thresholding is carried out on the edge image to eliminate false response, and a characteristic synthesis method is used to collect final edge information from multiple scales,
a uniform shift discontinuity preserving filtering step:
first, for each image pixel Xi, the initialization step number j is 1, Y(i,1)=XiSecond, calculate Y(i,j+1)Until convergence on Y(i,con)Finally defining the filtered pixel valuesNamely atThe filtered pixel value of (a) is assigned as the convergence pointI denotes the number of the pixel, ZiFor each image pixel;
mean shift image segmentation step:
first, mean shift discontinuity preserving filtering is adopted to preserve the convergence point of each d-dimensionAll information of, secondly all ZiClustering according to a kernel Hs in an airspace and a kernel Hr in a value range to obtain { Cp } p ═ 1i={p|Zi∈Cp1, n, and eliminating regions smaller than p pixels, where m is a natural number, Cp is the attraction field of the convergence point, and p and d represent the spatial dimensions of the image.
The intelligent camera hardware system provided by the invention has the beneficial effects that the intelligent camera hardware system takes ZYNQ as a core, and consists of a core board and an image acquisition board, wherein an image acquisition module consists of an optical component system, an image sensor, an AD conversion module and other components. The intelligent camera selects an embedded Linux operating system as a software platform, designs and develops intelligent camera configuration software by combining an OpenCV open source vision library, designs an image processing application program on an application layer to realize video data flow control of FPGA hardware processing and ARM cooperative processing, completes image processing of a large number of involved image processing algorithms, and realizes real-time acquisition and processing of high-definition video data streams.
Drawings
FIG. 1 is a system framework diagram of the present invention.
Detailed Description
The invention provides a small-sized wireless electrical and mechanical system based on 500 ten thousand CMOS chips and a ZYNQ system at the front end of an intelligent camera, which takes ZYNQ as a core and adopts an embedded Linux operating system, can directly process required image detection and measurement information in the camera, and transmits the information to external equipment through an RS232 serial port, Ethernet communication and input/output GPIO.
The intelligent camera provided by the invention has a small integral structure, integrates more than 40 image processing algorithm libraries such as horizontal mirror image, image zooming, image rotation, binarization, sub-pixel positioning and the like, can directly process and collect images internally, outputs results to control external equipment or outputs the results to other equipment through a serial port and a network port, and the results of visual detection and measurement of the intelligent camera are used by other equipment.
In the implementation process of the invention, the hardware of the intelligent camera adopts a low-power-consumption main chip xc7z020clg484 and a memory chip MT41J256M16, and integrates an Ethernet interface, a USB, a memory card, an RS232 serial port and an input/output GPIO into a whole. The intelligent camera completes the functions of an image acquisition module, an image preprocessing module and an image display module on the FPGA through the combination of a C/CS lens and a CMOS module, a Linux operating system is built on an ARM, the control function of the whole video acquisition processing flow is realized on an application layer, the idea of software and hardware cooperation is adopted in the image processing of the application layer, the acceleration of image preprocessing hardware is realized, and the whole image processing speed of the system is improved. The intelligent camera integrates 40 image processing algorithm libraries such as image template matching target searching, edge extraction and segmentation and the like on an application layer, and develops flexible image processing system software. The intelligent camera has the support of the hardware environment of the image processing unit, can directly complete the image processing function, can transmit an output result to other equipment through a serial port and Ethernet communication, and can directly control the output equipment through GPIO according to the processing result.
The following explains the intelligent camera integrated partial image processing algorithm provided by the invention:
1. target searching algorithm based on template matching
The algorithm is implemented as follows, and an input image is set as g (i, j).
(1) Image pre-processing
Wherein f (x, y) is the preprocessed image; θ is a local neighborhood of the current pixel (m, n). One-sided leave-on function existence inverse function u-1And a (i, j) is a weighting coefficient.
(2) Template matching
S (m, n) is the template image, M, N is the dimension of the template image, f (m, n) is a sub-image in the f (x, y) image that is the same size as the template image, and D (x, y) is a measure of matching error.
(3) Geometric transformation
And mapping the template image to the position of the processed image through geometric transformation.
T is a vector function. (x, y) are the pixel coordinates of the template image, and (x ', y') are the new coordinates of the transformed pixels in the processed image.
2. Edge extraction segmentation
(1) Edge extraction
The image is first convolved with a gaussian function of scale σ.
Next, for each pixel in the image, the normal n to the local edge is estimated.
The edge position is found again.
The edge strength is calculated again.
And finally, performing hysteresis thresholding on the edge image to eliminate false response, and collecting final edge information from multiple scales by using a feature synthesis method.
(2) uniform shift discontinuity preserving filtering
First, for each image pixel Xi, the initialization step number j is 1, Yi,1 is Xi, then Y (i, j +1) is calculated until convergence on Y (i, con), and finally the filtered pixel value is definedNamely atThe filtered pixel value of (a) is assigned as the convergence pointThe image value of the pixel of (1).
(3) Mean shift image segmentation
First, mean shift discontinuity preserving filtering is adopted to preserve the convergence point of each d-dimensionAll information of, secondly all ZiClustering according to a kernel Hs in an airspace and a kernel Hr in a value range to obtain { Cp } p ═ 1i={p|Zi∈Cp1.., n, and eliminating areas smaller than p pixels.
The intelligent camera hardware system provided by the invention has the beneficial effects that the intelligent camera hardware system takes ZYNQ as a core, and consists of a core board and an image acquisition board, wherein an image acquisition module consists of an optical component system, an image sensor, an AD conversion module and other components. The intelligent camera selects an embedded Linux operating system as a software platform, designs and develops intelligent camera configuration software by combining an OpenCV open source vision library, designs an image processing application program on an application layer to realize video data flow control of FPGA hardware processing and ARM cooperative processing, completes image processing of a large number of involved image processing algorithms, and realizes real-time acquisition and processing of high-definition video data streams.
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (3)
1. A smart camera based on CMOS chip and ZYNQ system is characterized by comprising,
ZYNQ platform, CMOS sensor;
the ZYNQ platform comprises an FPGA hardware unit and an ARM hardware unit;
the FPGA hardware unit comprises a video acquisition module and a video preprocessing module;
the video acquisition module acquires video signal information through the CMOS sensor and sends the video signal information to the video preprocessing module, and the video preprocessing module sends the processed video signal to the ARM hardware unit;
the ARM hardware unit runs a linux operating system, and an application layer video image processing module is installed in the linux operating system.
2. The CMOS chip and ZYNQ system based smart camera of claim 1, wherein said application layer video image processing module executes a template matching based target finding algorithm,
the template matching based object finding algorithm comprises the following steps,
an image preprocessing step:
wherein f (x, y) is the preprocessed image; theta is a local neighborhood of the current pixel (m, n), and the one-sided persistence function u has an inverse function u-1A (i, j) is a weighting coefficient, g (i, j) is an input image;
template matching:
s (m, n) is a template image, M, N is the dimension of the template image, f (m, n) is a sub-image in the f (x, y) image with the same size as the template image, and D (x, y) is the measure of matching error;
a geometric transformation step:
mapping the template image to the position of the processed image through geometric transformation,
t is a vector function, (x, y) is the pixel coordinates of the template image, and (x ', y') is the new coordinates of the transformed pixels in the processed image.
3. The CMOS chip and ZYNQ based smart camera of claim 1, wherein the application layer video image processing module executes an edge extraction segmentation algorithm,
the edge extraction segmentation algorithm comprises the following steps,
an edge extraction step:
the image is first convolved with a gaussian function of scale sigma,
secondly, for each pixel in the image, the normal n of the local edge is estimated,
the position of the edge is found again and,
the edge strength is calculated again and the edge strength is calculated,
finally, hysteresis thresholding is carried out on the edge image to eliminate false response, and a characteristic synthesis method is used to collect final edge information from multiple scales,
a uniform shift discontinuity preserving filtering step:
first, for each image pixel Xi, the initialization step number j is 1, Y(i,1)=XiSecond, calculate Y(i,j+1)Until convergence on Y(i,con)Finally defining the filtered pixel valuesNamely atThe filtered pixel value of (a) is assigned as the convergence pointI denotes the number of the pixel, ZiFor each image pixel;
mean shift image segmentation step:
first, mean shift discontinuity preserving filtering is adopted to preserve the convergence point of each d-dimensionAll information of, secondly all ZiClustering according to a kernel Hs in an airspace and a kernel Hr in a value range to obtain { Cp } p ═ 1i={p|Zi∈Cp1, n, and eliminating regions smaller than p pixels, where m is a natural number, Cp is the attraction field of the convergence point, and p and d represent the spatial dimensions of the image.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011251121.5A CN112351181A (en) | 2020-11-11 | 2020-11-11 | Intelligent camera based on CMOS chip and ZYNQ system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011251121.5A CN112351181A (en) | 2020-11-11 | 2020-11-11 | Intelligent camera based on CMOS chip and ZYNQ system |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112351181A true CN112351181A (en) | 2021-02-09 |
Family
ID=74363236
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011251121.5A Pending CN112351181A (en) | 2020-11-11 | 2020-11-11 | Intelligent camera based on CMOS chip and ZYNQ system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112351181A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114760414A (en) * | 2022-04-12 | 2022-07-15 | 上海航天电子通讯设备研究所 | Image acquisition and processing system for CMV4000 camera |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104820418A (en) * | 2015-04-22 | 2015-08-05 | 遨博(北京)智能科技有限公司 | Embedded vision system for mechanical arm and method of use |
CN105847766A (en) * | 2016-05-30 | 2016-08-10 | 福州大学 | Zynq-7000 based moving object detecting and tracking system |
CN206154352U (en) * | 2016-09-18 | 2017-05-10 | 常州机电职业技术学院 | Robot vision system and robot with motion object detection and tracking function |
CN110012201A (en) * | 2019-04-10 | 2019-07-12 | 山东尤雷克斯智能电子有限公司 | A kind of USB3.0 ultrahigh speed camera and its working method based on complete programmable SOC |
CN209748689U (en) * | 2019-03-06 | 2019-12-06 | 上海艾迪森国际数字医疗装备有限公司 | Image transmission device of double-network-port CMOS detector based on ZYNQ series FPGA + ARM |
-
2020
- 2020-11-11 CN CN202011251121.5A patent/CN112351181A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104820418A (en) * | 2015-04-22 | 2015-08-05 | 遨博(北京)智能科技有限公司 | Embedded vision system for mechanical arm and method of use |
CN105847766A (en) * | 2016-05-30 | 2016-08-10 | 福州大学 | Zynq-7000 based moving object detecting and tracking system |
CN206154352U (en) * | 2016-09-18 | 2017-05-10 | 常州机电职业技术学院 | Robot vision system and robot with motion object detection and tracking function |
CN209748689U (en) * | 2019-03-06 | 2019-12-06 | 上海艾迪森国际数字医疗装备有限公司 | Image transmission device of double-network-port CMOS detector based on ZYNQ series FPGA + ARM |
CN110012201A (en) * | 2019-04-10 | 2019-07-12 | 山东尤雷克斯智能电子有限公司 | A kind of USB3.0 ultrahigh speed camera and its working method based on complete programmable SOC |
Non-Patent Citations (3)
Title |
---|
张超: "树木影像特征提取与立体匹配技术研究", 《中国优秀博硕士学位论文全文数据库(博士)农业科技辑》 * |
胥泽飞: "基于Zynq的智能相机硬件设计", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 * |
金永涛等: "基于多特征的高分遥感图像分割算法研究", 《中国空间科学技术》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114760414A (en) * | 2022-04-12 | 2022-07-15 | 上海航天电子通讯设备研究所 | Image acquisition and processing system for CMV4000 camera |
CN114760414B (en) * | 2022-04-12 | 2024-04-16 | 上海航天电子通讯设备研究所 | Image acquisition and processing system for CMV4000 camera |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10198823B1 (en) | Segmentation of object image data from background image data | |
US11379699B2 (en) | Object detection method and apparatus for object detection | |
CN111665842B (en) | Indoor SLAM mapping method and system based on semantic information fusion | |
CN103886107B (en) | Robot localization and map structuring system based on ceiling image information | |
WO2020134818A1 (en) | Image processing method and related product | |
EP3376433B1 (en) | Image processing apparatus, image processing method, and image processing program | |
CN111191582B (en) | Three-dimensional target detection method, detection device, terminal device and computer readable storage medium | |
CN111444764A (en) | Gesture recognition method based on depth residual error network | |
CN112233221B (en) | Three-dimensional map reconstruction system and method based on instant positioning and map construction | |
CN115719436A (en) | Model training method, target detection method, device, equipment and storage medium | |
CN113743385A (en) | Unmanned ship water surface target detection method and device and unmanned ship | |
CN113223045A (en) | Vision and IMU sensor fusion positioning system based on dynamic object semantic segmentation | |
CN112351181A (en) | Intelligent camera based on CMOS chip and ZYNQ system | |
Esfahani et al. | DeepDSAIR: Deep 6-DOF camera relocalization using deblurred semantic-aware image representation for large-scale outdoor environments | |
Akman et al. | Multi-cue hand detection and tracking for a head-mounted augmented reality system | |
CN112884803A (en) | Real-time intelligent monitoring target detection method and device based on DSP | |
CN111967287A (en) | Pedestrian detection method based on deep learning | |
CN108058170A (en) | A kind of vision robot's data acquisition processing system | |
CN114200934A (en) | Robot target following control method and device, electronic equipment and storage medium | |
Shiratori et al. | Detection of pointing position by omnidirectional camera | |
CN112001247A (en) | Multi-target detection method, equipment and storage device | |
CN115147809B (en) | Obstacle detection method, device, equipment and storage medium | |
CN114419451B (en) | Method and device for identifying inside and outside of elevator, electronic equipment and storage medium | |
CN112634360B (en) | Visual information determining method, device, equipment and storage medium | |
CN110660134B (en) | Three-dimensional map construction method, three-dimensional map construction device and terminal equipment |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210209 |