CN105488780A - Monocular vision ranging tracking device used for industrial production line, and tracking method thereof - Google Patents

Abstract

The invention discloses a monocular vision ranging tracking device used for an industrial production line, and a tracking method thereof, and is applied to the industrial production line. The device comprises an image sensing unit, a control correction unit, a user graphical interface and a data storage module. The method comprises the following steps: 1) obtaining a defocusing blurred image through an image acquisition module, and establishing a mathematical model; 2) adopting a cepstrum algorithm to operate the mathematical model, separating true input image information and the information of a defocusing degenerate kernel function; 3) analyzing the relevant information of the defocusing degenerate kernel function to quantificationally obtain a point diffusion radius; and 4) according to the point diffusion radius, obtaining the distance information of a feature capturing object, and combining with the 2D (Two-Dimensional) positioning information of the feature capturing object to finally obtain a 3D (Three-Dimensional) positioning coordinate of the feature capturing object. The monocular vision ranging tracking device has the advantages that system cost is greatly lowered since only one midrange industrial camera is used, and points do not need to be subjected to initial calibration and labeling.

Description

A monocular vision ranging tracking device and tracking method for industrial production lines

technical field

The invention belongs to the technical field of industrial robots, and in particular relates to a monocular vision ranging tracking device and a tracking method for industrial production lines.

Background technique

In recent years, with the in-depth understanding of the functional structure of the brain's visual cortex (VisualCortex), the emerging discipline of computer vision has also received more and more attention, and has made considerable and in-depth development. Visual measurement and 3D motion positioning and tracking technologies are also widely used in many corresponding fields, such as: motion tracking games based on positioning control in the entertainment game industry; the military industry benefits from visual real-time positioning technology to identify and capture; in the medical industry, human-computer interaction needs to be simulated; in artificial intelligence, vision is used to control robotic systems and other fields.

At present, most ranging and positioning systems use binocular vision, which is based on the theory of binocular stereo vision, which is based on the research on the human visual system and obtains scene information through binocular parallax. Marr, Poggio and Grimson first proposed and implemented a computational vision model and algorithm based on the human visual system. Although binocular vision is the 3D data recovery method closest to human vision, in fact, there are certain problems in the traditional binocular positioning system, such as: traditional binocular positioning system covers the blind area, the feature point matching process is complicated, and high precision is required binocular collaboration.

Contents of the invention

The purpose of the present invention is to overcome the existing binocular vision distance measurement and tracking system: the phenomenon of occlusion of blind areas, the need for complex feature point matching processes in the execution of distance measurement algorithms and the need for high-precision binocular cooperation in control operations. A series of problems, a monocular vision ranging tracking device and tracking method for industrial production lines are proposed.

A monocular vision ranging tracking device for industrial production lines, including: an image sensing unit, a control and correction unit, a graphical user interface and a data storage module;

The image sensing unit includes an image acquisition module and an image preprocessing module for image acquisition.

Among them, the image acquisition module is the hardware acquisition part of the image sensing unit, which collects the image of the feature capture object, and outputs the collected original image information to the image preprocessing module; the image preprocessing module is the software algorithm processing part, which utilizes the image The preprocessing algorithm performs image processing on the original image information: illumination normalization processing, white balance processing, exposure control processing and image convolution; the processed image information is output to the control correction unit.

The control and correction unit includes a computer controller and an execution terminal module, which is used to receive the image information output by the image sensing unit, and control the execution terminal module to perform corresponding operations according to corresponding user instructions.

The computer controller receives the image information output by the image preprocessing module, and receives the user instruction obtained from the user graphical interface interface, processes the image and extracts and mines the data according to the instruction, and generates a control signal from the image processing data in the form of a control flow Output to the execution terminal block. In addition, the computer controller also stores the image processing data into the corresponding data storage module according to the user's instruction.

The execution terminal module is connected to the image acquisition module, and receives the image processing data output by the computer controller for correction and adjustment.

The user graphical interface interface transmits user instructions to the computer controller, and at the same time receives image processing instructions and operation instructions required by the user from the computer controller;

The data storage module stores the image processing data output by the computer controller.

A monocular vision ranging tracking method for industrial production lines, the specific steps are as follows:

Step 1, the user graphical interface interface outputs user instructions to the computer controller, and controls the image acquisition module on the execution terminal module to perform image acquisition on the feature capture object;

Step 2, the image preprocessing module uses the image preprocessing algorithm to process the collected image, and outputs the image information to the computer controller;

Step 3, the computer controller calculates the distance between the feature capture object and the execution terminal module by processing the image and data extraction and mining operations, so as to realize monocular vision distance measurement and tracking;

The method for the computer controller to process images and extract and mine data, the specific steps are as follows:

Step 1: Obtain a defocused blurred image according to the image acquisition module, and establish a mathematical model.

The mathematical model of the formation process of the defocused blurred image is:

f(ε,η)*h(x,y)+n(x,y)=g(x,y)

Where f(ε,η) is the real input defocus blur image obtained on the focus plane, h(x,y) is the defocus degradation kernel function, n(x,y) is the environmental noise, g(x,y) is The actual output defocused blurred image;

Among them, the defocus degradation kernel function adopts the point spread function model, that is, the circular point spread degradation function form:

r is the point diffusion radius of the degenerate kernel function, which quantitatively illustrates the blurring degree of the defocused image; x and y are the two-dimensional coordinate values in the Cartesian coordinate system of the corresponding diffusion point in the image.

Step 2: Use the cepstrum algorithm to operate the mathematical model to separate the real input image information from the information of the defocus degradation kernel function.

Step 201, express the actual output defocus blurred image information data as: the convolution form of the real input image information data and the defocus degradation kernel function;

g(x,y)=f(x,y).*h(x,y)

Step 202, using Fourier transform to convert the convolution result in step 201 into a multiplication form in the frequency domain;

DFT(f(x,y).*h(x,y))=DFT(f(x,y))*DFT(h(x,y))

Step 203, utilizing the properties of the logarithmic function, converting the multiplication result in the frequency domain in step 202 into an addition form;

log(DFT(f(x,y)))+log(DFT(h(x,y)))

Step 204, the addition form separates the real input image information from the information of the defocus degradation kernel function;

Step 205, perform inverse Fourier transform again on the result obtained in step 204, and complete the cepstrum algorithm for the image.

The inverse Fourier transform is performed on the logarithmic frequency domain image processing result obtained in step 204 again to complete the cepstrum algorithm for the image, thereby converting it into the cepstrum domain, and analyzing its characteristic information in the cepstrum domain.

Step 3: Through the information related to the defocus degradation kernel function separated in step 2, the point diffusion radius indicating the degree of blur of the defocused image is obtained, further quantitatively.

Step 301, three-dimensionally display the data acquired through the cepstrum algorithm in step 2;

Step 302, according to the envelope result of the three-dimensional image on the two-dimensional plane, the radius of the ring groove closest to the center point is obtained;

Step 303, the obtained ring groove radius is proportional to the point diffusion radius r.

Step 4: Obtain the distance information of the feature capture object through step 3 combined with the 2D positioning of the feature capture object, and finally obtain the 3D positioning coordinates of the feature capture object.

The advantages of the present invention are:

(1) A monocular vision ranging tracking method for industrial production lines, which does not require initialization calibration and marking points, and only needs to ensure that the object is an opaque body with uniform coloring.

(2) A monocular vision distance measurement and tracking device for industrial production lines, which only uses a single mid-range industrial camera, which greatly reduces hardware costs and eliminates the redundancy of the traditional binocular vision distance measurement and positioning device described in the background above. The advantages and disadvantages are eliminated, so that distance measurement and tracking can be performed simply and conveniently.

(3) A monocular vision ranging tracking method for industrial production lines, which effectively removes the influence of environmental factors such as light and micro vibration on the system.

Description of drawings

Fig. 1 is a schematic diagram of a monocular vision distance measuring and tracking device used in an industrial production line according to the present invention;

Fig. 2 is a schematic diagram of the image sensing unit in the monocular vision ranging tracking device of the present invention;

Fig. 3 is a schematic diagram of a mathematical model of a defocused blurred image of the present invention;

Fig. 4 is a schematic diagram of the distance information of the object to be quantitatively calculated and captured by the present invention;

Fig. 5 is a principle diagram of monocular vision distance measurement and tracking in the present invention;

Fig. 6 is a flow chart of a monocular vision distance measurement and tracking method used in an industrial production line according to the present invention;

Fig. 7 is the flow chart of the method for image processing and data extraction and mining operation by the computer controller of the present invention;

Fig. 8 is a flow chart of the 2D positioning algorithm in the monocular vision ranging tracking method of the present invention;

detailed description

The specific embodiments of the present invention will be further described below in conjunction with the accompanying drawings.

A monocular vision ranging and tracking device for industrial production lines is applied to the ranging and tracking of industrial robot motion operations on industrial production lines. As shown in Figure 1, it includes: an image sensing unit, a control correction unit, a graphical user interface and a data storage module;

The image sensing unit includes an image acquisition module and an image preprocessing module for image acquisition;

Among them, the image acquisition module is the hardware acquisition part of the image sensing unit, which uses the image acquisition equipment to acquire images of the feature capture objects, and outputs the acquired original image information to the image preprocessing module through the image data bus for the following One step image preprocessing step.

The image acquisition device is preferably a camera, video camera or network camera; in this embodiment, a megapixel-level industrial fixed-focus camera is used, which has an imaging distortion correction function. Among them, the industrial fixed-focus camera adopts a fixed focal length lens. The parameters of the lens are shown in Table 1. The processor core is Linux3.0 kernel and above, the control operating system is Ubuntu11.04 and above, and has corresponding GigE Gigabit Ethernet interface.

Table 1 is the fixed focal length lens parameters of industrial fixed focus cameras

lens name data focal length 12.00mm aperture F1.4-16C hardware interface GigE Gigabit Ethernet camera image plane 1024*1280

The image preprocessing module is the software algorithm processing part of the image sensing unit.

For any existing image acquisition device, a set of software package (SDK) is required to support it. The vision software package matched with the acquisition device usually includes the following modules: image preprocessing part; image processing part; character recognition part; data Extraction part; image resource management part; display function part and other function part.

As shown in Figure 2, the image acquisition module uses the fixed-focus lens of the industrial fixed-focus camera and GigE Gigabit Ethernet to collect the image of the feature-captured object and transmits it to the computer controller; at the same time, the image pre-processing module uses the industrial fixed-focus camera to automatically With the image preprocessing algorithm in the software development kit SDK, it processes the original image information output by the image acquisition module, including: setting the image capture mode, initializing the image quality, and initial simple image preprocessing functions; the simple image processing process , including: illumination normalization processing, white balance processing, exposure control processing and image convolution processing;

Illumination normalization processing: remove the influence of different illumination conditions on subsequent image processing;

White balance processing: perceive the diameter of the tube around the image acquisition, adjust the color balance, and correct and adjust the image signal output by the image acquisition module;

Exposure control processing: adjust the overall brightness of the image;

Image convolution: performs edge processing on images.

The specific processing steps of the image preprocessing module are as follows:

The first step is to set Gamma correction to improve the influence of light on the quality of the acquired image;

According to the Gammacorrection function in the AnalogControls function control area of the industrial fixed-focus camera, set GammaEnable to allow the camera to automatically perform Gamma correction on the image when acquiring the image and performing preprocessing.

The second step is to set the automatic white balance, and adjust the white balance according to the lighting conditions under different environmental conditions;

The function of setting white balance in the AnalogControls function control area of industrial fixed-focus cameras is divided into two types: manual setting and automatic setting. In manual setting mode: select the required white balance value by setting BalanceRatio; in automatic setting mode, select The BalanceWhiteAuto option enables the camera to automatically perform white balance adjustments when capturing images for image preprocessing.

The third step is to set the automatic exposure, and adjust the optimal exposure time according to the light intensity under different environmental conditions.

The AcquisitionControls function control area that comes with the industrial fixed-focus camera encapsulates the function of setting the exposure mode/time/time base, which is divided into manual setting and automatic setting. In the manual setting mode, by setting ExposureMode/Time/Timebase, etc. Option settings to adjust the exposure parameters required by the user; in the automatic setting mode, select the ExposureAuto option to enable the camera to automatically set the optimal exposure parameters according to the environment when capturing images.

Finally, the preprocessed image with less error and noise after the above image preprocessing is output to the control correction unit of the next layer.

The control and correction unit includes a computer controller and an execution terminal module, which is used to receive the image information output by the image sensing unit, and control the execution terminal module to perform corresponding operations according to corresponding user instructions.

The computer controller obtains the image information package output by the image preprocessing module, and receives user instructions obtained from the user graphical interface interface, performs specific post-image processing and data extraction and mining operations on the image information package according to the instructions, and generates corresponding control The signal is output to the execution terminal module through the data bus in the form of control flow.

In addition, the computer controller also stores the data that the user is interested in according to the user's instruction, and stores the data in the corresponding data storage module, so that the user can call the data in the data storage module for viewing and other functions when needed. Processing operations.

The execution terminal module receives the correction and control information output by the computer controller, and corrects and adjusts the posture information such as the posture and position of the terminal, so as to finally achieve the operation purpose specified by the user.

The graphical user interface belongs to the auxiliary part, which is used to generate a friendly human-computer interaction interface; on the one hand, the user instruction is sent to the computer controller, and at the same time, the image processing instruction and operation instruction required by the user are received from the computer controller;

The data storage module belongs to the auxiliary part and is used for storing the real-time data output by the computer controller.

A monocular vision distance measurement and tracking method used in industrial production lines is applied to the production line in the field of industrial control; the principle of monocular vision distance measurement and tracking, as shown in Figure 5, uses the physical optical imaging principle of the defocus lens imaging system , where the image acquisition device is simplified as a fixed-focus lens. Due to the different distances, the feature capture objects cannot be accurately focused on the clear focus plane, resulting in a certain degree of defocus diffusion degradation. The actual output image information obtained has a certain degree of defocusing point diffusion effect, and the quantitative expression of the defocusing point diffusion effect is the point diffusion radius, which is determined by the physical optics lens imaging rule formula:

$\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; u</span>}}<span\; class="notranslate">\; +</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; v</span>}}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; f</span>}}$

Among them, u is the object distance, which is the distance between the feature capture object and the lens; f is the focal length of the lens; v is the focusing imaging image distance.

From the ratio of similar triangles, it can be seen that there is a quantitative relationship between the point diffusion radius and the object distance u: when the feature capture object is clearly imaged on the focal plane, the actual imaging image distance is v ₁ before the clear focal plane, and the point diffusion radius is R ₁ ;

$\frac{{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}}{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; v</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}}{\mathrm{<span\; class="notranslate">\; v</span>}}$

After the clear focus plane, the actual imaging image distance is v ₂ , and the point diffusion radius is R ₂ ;

$\frac{{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}}{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; v</span>}}{\mathrm{<span\; class="notranslate">\; v</span>}}$

Among them, d is the lens diameter;

According to the point diffusion radius, the distance u between the feature capturing object and the lens is obtained, and the ranging tracking of the capturing object is performed.

The software algorithm is based on the system idea of modern control methods in informatics theory, and the obtained blurred image is set to be a clear image above the focus plane through the blurring effect of the defocus degradation kernel function, which also needs to consider the imaging process. exposure to ambient noise.

As shown in Figure 6, the specific steps of the monocular vision ranging tracking method are as follows:

Step 1, the user graphical interface interface outputs user instructions to the computer controller, and controls the image acquisition module on the execution terminal module to perform image acquisition on the feature capture object;

Step 2, the image preprocessing module uses the image preprocessing algorithm to process the collected image, and outputs the image information to the computer controller;

Step 3, the computer controller calculates the distance between the feature capture object and the execution terminal module by processing the image and data extraction and mining operations, so as to realize monocular vision distance measurement and tracking;

The method for image processing and data extraction mining operation by the computer controller is as shown in Figure 7, and the specific steps are as follows:

Step 1: Obtain a defocused blurred image according to the image acquisition module, and establish a mathematical model.

The mathematical model is based on the idea of image degradation, that is, the actual input image is formed by the image acquisition module to form the actual output image. Due to the physical properties of the image acquisition module itself and the influence of environmental noise, the actual input image is not the same as the actual output image.

As shown in Figure 3: the mathematical model of the formation process of defocus blur image is:

f(ε,η)*h(x,y)+n(x,y)=g(x,y)

Where f(ε,η) is the real input image obtained on the focus plane, h(x,y) is the defocus degradation kernel function, n(x,y) is the environmental noise, g(x,y) is the actual output Defocus blurred image.

Among them, the defocus degradation kernel function adopts the point spread function model, that is, the circular point spread degradation function form:

r is the point diffusion radius of the degenerate kernel function, which quantitatively illustrates the blurring degree of the defocused image; x and y are the two-dimensional coordinate values in the Cartesian coordinate system of the corresponding diffusion point in the image.

The control correction unit receives the defocused blurred image obtained by the image acquisition module, and combines the user command correction and control information obtained from the user graphical interface interface to correct and adjust the pose information of the terminal such as posture and position, and finally achieve the operation purpose specified by the user .

Step 2: Use the cepstrum algorithm to operate the mathematical model to separate the real input image information from the information of the defocus degradation kernel function.

The specific steps to obtain the point diffusion radius r based on the cepstrum algorithm are as follows:

Step 201, express the actual output defocus blurred image information data as: the convolution form of the real input image information data and the defocus degradation kernel function;

g(x,y)=f(x,y).*h(x,y)

Step 202, using Fourier transform to convert the convolution result in step 201 into a multiplication form in the frequency domain;

DFT(f(x,y).*h(x,y))=DFT(f(x,y))*DFT(h(x,y))

Step 203, utilizing the properties of the logarithmic function, converting the multiplication result in the frequency domain in step 202 into an addition form;

log(DFT(f(x,y)))+log(DFT(h(x,y)))

Step 204, the addition form separates the real input image information from the information of the defocus degradation kernel function;

Step 205, perform inverse Fourier transform again on the result obtained in step 204, and complete the cepstrum algorithm for the image.

The inverse Fourier transform is performed on the logarithmic frequency domain image processing result obtained in step 204 again to complete the cepstrum algorithm for the image, thereby converting it into the cepstrum domain, and analyzing its characteristic information in the cepstrum domain.

The data storage module stores the real input image information and the information of the defocus degradation kernel function, and interacts with the user graphical interface.

Step 3: By analyzing the relevant information of the defocus degradation kernel function separated in step 2, a point diffusion radius indicating the degree of blur of the defocused image is obtained.

Step 301, three-dimensionally display the data acquired through the cepstrum algorithm in step 2;

The data were processed using OpenCV software.

Step 302, according to the result of the envelope of the three-dimensional image on the two-dimensional plane, the radius of the ring groove closest to the center point is obtained;

The two-dimensional plane refers to the X-Z/Y-Z plane.

Step 303, the obtained ring groove radius is proportional to the point diffusion radius r.

As shown in Figure 4, the ordinate is set to y=0 as the zero plane in the cepstrum domain, referred to as the zero line, and x≈641 is the center peak of the cepstrum domain processing result image; different curves represent those obtained by using the cepstrum algorithm, The two-dimensional image of the processing result of the defocus degradation kernel function with different r, the intersection point of the defocus degradation kernel function curve and the zero line of different r is the zero point, and the horizontal distance from the intersection point closest to the central peak x≈641 of the zero point to the central peak value is The value of the groove radius, which is proportional to the point spread radius r.

In the present invention, the radius of the annular groove is twice the point diffusion radius r.

So far, the point diffusion radius r is expressed quantitatively by the radius of the ring groove, so as to further quantitatively calculate the distance information of the captured object.

The data storage module stores the distance information of the captured object and interacts with the graphical user interface.

Step 4: Obtain the distance information of the feature capture object through step 3 combined with the 2D positioning of the feature capture object, and finally obtain the 3D positioning coordinates of the feature capture object.

For 2D positioning of feature-captured objects, an automatically correctable extraction algorithm is used. The 2D positioning algorithm performs automatic relocation function, as shown in Figure 8, and the specific steps are as follows:

Step 401, setting a Poisson kernel function to preprocess the original acquired image;

Step 402, tracking the feature capture object set by initialization, and obtaining the point diffusion radius according to the feature capture object;

Step 403, acquiring continuous image correlation and difference information in the time domain;

Step 404, according to the image correlation and difference information, judge whether the continuous images are stable, if stable, go to step 405, otherwise go to step 406;

Step 405, according to the stable continuous image, confirm the point diffusion radius, and obtain the 3D positioning coordinates of the captured feature object;

According to the stable continuous image, confirm the point diffusion radius and output it to the user graphical interface, capture the feature object, and obtain the 3D positioning coordinates of the feature capture object;

Step 406 , if the acquired image is highly unstable, then the point diffusion radius of the feature capturing object is wrong, and return to step 402 .

The software algorithm part uses the idea of monocular close-range distance measurement, based on the distance measurement algorithm of the motion defocus blur algorithm, and reflects the distance information of the feature capture object through the blur degree of a series of defocus blur images obtained. At the same time, applying informatics theory, based on the frequency domain cepstrum algorithm, the point diffusion radius of the point diffusion kernel function of the defocused blurred image is obtained, and then the degradation system function of the image blur degradation model is obtained, so as to accurately and quantitatively calculate the The distance information of the feature capture object is finally supplemented with the corresponding 2D automatic correction positioning algorithm to obtain the distance measurement and positioning coordinate information of the feature capture object.

Since the system uses a new algorithm processing idea and is mainly used in the field of industrial robots, it must consider its special requirements and reliability issues in the field of industrial robots. The operating system of an industrial robot requires less fluctuation and higher stability. Based on the above reasons, a smoothing filter algorithm is added to the algorithm of the present invention to provide relatively smooth and stable positioning data for the controller.

Claims (5)

1. A monocular vision ranging tracking device for an industrial production line, comprising an image sensing unit, a control correction unit, a graphical user interface and a data storage module; The image sensing unit includes an image acquisition module and an image preprocessing module. The image acquisition module collects the image of the feature capture object and outputs it to the image preprocessing module; the image preprocessing module uses the image preprocessing algorithm to perform image processing on the image information, and output to the control correction unit; The control correction unit includes a computer controller and an execution terminal module. The computer controller receives the image information output by the image preprocessing module, and receives the user instruction output by the user graphical interface interface, and processes the image and performs data extraction and mining operations according to the instruction. Process the data to generate a control signal, output it to the execution terminal module, and control the execution terminal module to perform corresponding operations; at the same time, the computer controller stores the image processing data in the data storage module according to the user's instruction; The execution terminal module is connected to the image acquisition module, and receives the image processing data output by the computer controller for correction and adjustment; The user graphical interface interface transmits user instructions to the computer controller, and at the same time receives the image processing instructions and operation instructions required by the user; The data storage module stores the image processing data output by the computer controller. 2. A monocular vision ranging and tracking device for industrial production lines according to claim 1, wherein the image acquisition device is a megapixel industrial fixed-focus camera. 3. A kind of monocular vision ranging tracking device for industrial production line as claimed in claim 1, it is characterized in that, described image preprocessing module carries out following operation to image information: set image capturing mode, initialize image quality And image preprocessing; image preprocessing includes: illumination normalization processing, white balance processing, exposure control processing and image convolution processing. 4. A kind of monocular vision ranging tracking method for industrial production line of application claim 1, is characterized in that, concrete steps are as follows: Step 1, the user graphical interface interface outputs user instructions to the computer controller, and controls the image acquisition module on the execution terminal module to perform image acquisition on the feature capture object; Step 2, the image preprocessing module uses the image preprocessing algorithm to process the collected image, and outputs the image information to the computer controller; Step 3, the computer controller calculates the distance between the feature capture object and the execution terminal module by processing the image and data extraction and mining operations, so as to realize monocular vision distance measurement and tracking; The method for the computer controller to process images and extract and mine data, the specific steps are as follows: Step 1: Obtain a defocused blurred image according to the image acquisition module, and establish a mathematical model; The mathematical model of the formation process of the defocused blurred image is: f(ε,η)*h(x,y)+n(x,y)=g(x,y) Where f(ε,η) is the real input defocus blur image obtained on the focus plane, h(x,y) is the defocus degradation kernel function, n(x,y) is the environmental noise, g(x,y) is The actual output defocused blurred image; Among them, the defocus degradation kernel function adopts the point spread function model, that is, the circular point spread degradation function form: r is the point diffusion radius of the degenerate kernel function; x and y are the two-dimensional coordinate values in the Cartesian coordinate system corresponding to the diffusion point in the image; Step 2: Use the cepstrum algorithm to operate the mathematical model to separate the real input image information from the information of the defocus degradation kernel function; Specifically include: Step 201, express the actual output defocus blurred image information data as: the convolution form of the real input image information data and the defocus degradation kernel function; g(x,y)=f(x,y).*h(x,y) Step 202, using Fourier transform to convert the convolution result in step 201 into a multiplication form in the frequency domain; DFT(f(x,y).*h(x,y))=DFT(f(x,y))*DFT(h(x,y)) Step 203, utilizing the properties of the logarithmic function, converting the multiplication result in the frequency domain in step 202 into an addition form; log(DFT(f(x,y)))+log(DFT(h(x,y))) Step 204, the addition form separates the real input image information from the information of the defocus degradation kernel function; Step 205, performing an inverse Fourier transform on the result obtained in step 204 again to complete the cepstrum algorithm for the image; Inverse Fourier transform is performed again on the log frequency domain image processing result obtained in step 204, and the cepstrum algorithm to the image is completed, thereby converting it into the cepstrum domain, and analyzing its characteristic information in the cepstrum domain; Step 3: Through the defocus degradation kernel function information separated in step 2, the point diffusion radius indicating the blur degree of the defocused image is obtained; Display the separation information obtained by the cepstrum algorithm in step 2 in three dimensions, and obtain the radius of the annular groove closest to the center point through the envelope result of the three-dimensional image on the two-dimensional plane, and the radius of the annular groove is proportional to the point diffusion radius r , the point diffusion radius r represents the distance information of the feature capture object; Step 4: Obtain the distance information of the feature capture object through step 3 combined with the 2D positioning of the feature capture object, and finally obtain the 3D positioning coordinates of the feature capture object. 5. A kind of monocular vision ranging tracking method for industrial production line as claimed in claim 4, is characterized in that, the specific steps of 2D positioning described in step 3 are as follows: Step 401, setting a Poisson kernel function to preprocess the original acquired image; Step 402, tracking the feature capture object set by initialization, and obtaining the point diffusion radius according to the feature capture object; Step 403, acquiring continuous image correlation and difference information in the time domain; Step 404, according to the image correlation and difference information, judge whether the continuous images are stable, if stable, go to step 405, otherwise go to step 406; Step 405, according to the stable continuous image, confirm the point diffusion radius, and obtain the 3D positioning coordinates of the captured feature object; Step 406 , if the acquired image is highly unstable, then the point diffusion radius of the feature capturing object is wrong, and return to step 402 .