CN115268012A - Real-time automatic focusing method of camera based on edge tracking - Google Patents

Abstract

The invention relates to a real-time automatic focusing method for cameras based on edge tracking, and belongs to the technical field of machine vision. First, a vision system is established. The vision system includes a light source, an imaging unit, a focusing unit, and the like, and the imaging unit includes a first camera and a second camera. , the focusing unit includes a moving platform and a mirror, the first camera is fixed on the moving platform, and the moving platform is driven by a motor or moves backward, so as to achieve the purpose of focusing; the second camera images through the mirror, when the object moves back and forth, the object The imaging of the second camera will move up and down, thereby generating displacement, so automatic focusing can be realized by establishing the positional relationship between the displacement and the motor. The automatic focusing method of the invention can avoid time-consuming definition calculation, has small calculation workload, fast automatic focusing speed, can meet the production needs of real-time focusing, and at the same time does not need to use an expensive zoom lens, and has low production cost.

Description

A Real-time Automatic Focusing Method of Camera Based on Edge Tracking

technical field

The invention belongs to the technical field of machine vision, and in particular relates to a camera real-time automatic focusing method based on edge tracking.

Background technique

In the industrial field, a clear image can more accurately reflect the object to be observed and obtain more accurate information. However, in actual application scenarios, the object to be observed will move back and forth relative to the imaging system, causing the image to be sometimes clear (that is, in-focus state) and sometimes blurred (that is, out-of-focus state). The traditional focusing method mainly relies on human experience to manually adjust the focus to obtain the ideal imaging effect, but requires separate training for practitioners, requires high labor costs, and has disadvantages such as strong subjectivity, low efficiency, and poor real-time performance. At the same time, the industrial production environment has the characteristics of high real-time, fast speed, high complexity, and poor security, which also make it difficult for humans to do it.

Therefore, it is necessary to use automatic focusing to meet the needs of industrial production. There are also related technical disclosures in the prior art. For example, in the Chinese invention patent application CN104570568A (automatic focusing method for projectors), it is necessary to repeat the focusing many times during the operation. It is difficult to meet the real-time production needs in the production process; in the Chinese invention patent application CN104917970A (an automatic focusing method), it is necessary to perform complex FFT transformation on each picture during the focusing process to obtain high-frequency components and low-frequency components. Further estimation of sharpness needs to consume a large amount of computing resources, and it is also difficult to meet the real-time production needs in the production process.

Contents of the invention

The purpose of the present invention is to provide a real-time automatic camera focusing method based on edge tracking to solve the technical problems raised in the background technology.

In order to achieve the above object, the present invention discloses a real-time automatic focusing method of a camera based on edge tracking. First, a visual system is built, and then a positive focus lookup table is constructed and real-time focusing is performed;

The constructing the positive focus lookup table includes, in the offline state, determining the focus range of the camera in the vision system, and constructing the positive focus lookup table;

The real-time focusing includes, in the online state, in the vision system, placing the object to be measured in the field of view of the camera, first driving the camera to move from the zero point to the maximum stroke, and collecting pictures to obtain the edge of the focus The center position of the positive focus lookup table is traversed to determine the nearest point as the reference point; then the picture is collected in real time and the absolute value of the difference between the center of the picture edge and the reference point is calculated, and then the positive focus lookup table is traversed to find the point that is the same as the current picture The closest point to the center of the edge, get the camera stroke at that point, and drive the camera to the stroke to achieve real-time focusing.

Further, the vision system includes an optical unit, an imaging unit, and a focusing unit; the imaging unit includes a first camera and a second camera; the focusing unit includes a first motion platform, a second motion platform, and a mirror; The first camera is fixed on the first moving platform, the second camera is fixed on one side of the first moving platform; the object to be measured is fixed on the second moving platform; the optical unit is fixed on the object to be measured The mirror is fixed on the other side of the object to be measured, and the mirror is located in the field of view of the second camera; the moving direction of the second moving platform is parallel to the optical axis of the first camera.

Further, in the process of constructing the positive focus lookup table,

First determine the focusing range of the first camera, including: set the stroke of the first motion platform to zero, drive the second motion platform to reciprocate, and find the positive focus A; set the stroke of the first motion platform to the maximum, and drive the second motion The platform reciprocates to find the positive focus B, and obtain the focusing range (A, B) of the first camera;

Then constructing the positive focus lookup table includes: dividing the focusing range of the first camera into equal parts to construct the positive focus lookup table.

Further, the focusing range of the first camera is equally divided according to the formula M=1+round((B-A)/step1);

Wherein, round() is a rounding operation, and step1 is 1/5 to 1/2 of the depth of field of the first camera; the positive focus lookup table includes M elements, and each element includes the first motion platform in positive focus. The position of the center of stroke and edge.

Further, the step1 is 1/3 of the depth of field of the first camera.

Further, in the process of constructing the positive focus lookup table, when determining the stroke of the first motion platform and the position of the edge center at the time of positive focus, it includes:

First, drive the second motion platform to position A, move the first motion platform from the zero point to the maximum stroke, and in the process, the first camera and the second camera simultaneously collect pictures, and binarize the image of the second camera , then detect the edge of the object to be tested, then fit the straight line of the edge, and finally calculate the center of the edge on the image;

Then calculate the sharpness D of the image of the first camera, find the maximum sharpness D _max , and record the center of the edge and the travel position of the first moving platform at this time in the positive focus lookup table, and observe whether the second moving platform moves to If it is point B, it ends, if not, it continues to drive the second motion platform forward until it moves to point B.

Wherein, when looking for the maximum sharpness, sharpness evaluation functions can be utilized, such as spatial domain methods, frequency domain methods, wavelet transform methods and information statistics methods, etc., in the present invention, use spatial domain methods to pass statistical image gradient magnitudes (such as Sobel transform, Laplace transform, etc.), because the positive focus has the clearest image, that is, has the strongest gradient value.

Further, the straight line of the edge is recorded as y=k*x+b, and finally the center (x _c , y _c ) of the edge on the image is calculated, and the calculation formula is as follows:

x _c = (Width-1)*0.5;

y _c =k*x _c +b;

Wherein, Width is the width of the image of the second camera.

Further, the method for binarizing the image of the second camera includes a single threshold method and/or an OSTU algorithm.

Further, the method for detecting the edge of the object to be measured includes Canny operation and/or Sobel operation.

Further, during the real-time focusing process,

First find the reference point: drive the first motion platform to move from zero point to the maximum stroke, the first camera and the second camera collect pictures, obtain the center (x _c , y _c ) of the edge of the positive focus, and then traverse the positive focus lookup table, Find the closest point and record it as the reference point R(x,y,Pos);

Then perform real-time focusing: the first camera collects pictures, calculates the center of the edge (x _c , y _c ), and calculates the absolute value of the difference between y _c and the y component of the reference point R d=abs(y _c -y), when When d>ste1p, then traverse the positive focus lookup table to find the closest point to the center (x _c , y _c ) of the current edge, obtain the stroke of the first motion platform at this point, and then drive the first motion platform to drive the first camera to With this stroke, real-time focusing can be realized.

In the present invention, the imaging unit includes a first camera and a second camera, wherein the first camera is fixed on the first moving platform, and the first moving platform can move forward or backward, so as to achieve the purpose of focusing; the second camera forms an image through a mirror , when the object to be measured moves back and forth, the image of the object to be measured on the second camera will move up and down, resulting in a displacement, so the automatic focusing can be realized by establishing the positional relationship between the displacement and the first motion platform.

Compared with the prior art, the camera's real-time automatic focusing method based on edge tracking of the present invention has the following advantages:

(1) In the present invention, the process of constructing the positive focus lookup table can be implemented offline, and the operation is convenient.

(2) In the present invention, in the real-time focusing process, the first motor is driven in real time through the positive focus lookup table to control the first motion platform to move back and forth so as to realize real-time automatic focusing. This method can avoid time-consuming sharpness calculations, It can also avoid the use of expensive zoom lenses.

(3) The automatic focusing method of the present invention is controlled by an algorithm, does not need to repeat focusing many times, has a small computational workload and high speed, and can meet the production needs of real-time focusing.

Description of drawings

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

Fig. 1: Schematic diagram of the structure of the vision system in Example 1.

Fig. 2: Working steps of constructing a positive focus lookup table in Example 1.

Fig. 3: The working steps of determining the stroke of the first motor and the position of the edge center in the first embodiment in positive focus.

Figure 4: Working steps of real-time focusing in Example 1.

Among them: 1. Light source; 2. First camera; 3. First lens; 4. Second camera; 5. Second lens; 6. First moving platform; 7. First motor; 8. Second moving platform; 9. The second motor; 10. The object to be measured; 11. The mirror.

Detailed ways

The technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings.

Example 1

The present invention mainly solves the problem of difficult high-quality imaging in industrial application scenarios. By pre-recording the corresponding relationship between the imaging position of the object edge at the second camera 4 and the position of the first motor 7 when the first camera 2 is in focus, a positive focus search is constructed. table; then in practical applications, the first motor 7 is driven in real time through the positive focus lookup table to control the first motion platform 6 to move back and forth so as to realize real-time automatic focusing. This method can not only avoid time-consuming sharpness calculation, but also avoid the use of expensive zoom lenses.

The first camera 2 is fixed on the first moving platform 6 (the advance or retreat of the first moving platform 6 is driven by the first motor 7), and the texture information of the calculation definition is provided for finding the positive focus; The mirror 11 forms an image and tracks the front and rear changes of the object 10 to be measured, so that the relationship between the displacement and the position of the first motor 7 can be established to realize automatic focusing.

A real-time automatic focusing method of a camera based on edge tracking is carried out under a specific vision system. First, a vision system needs to be built. As shown in Figure 1, the vision system includes an optical unit, an imaging unit, a focusing unit, a computer and a control unit. unit, etc., wherein the optical unit is the light source 1, the imaging unit includes the first camera 2, the first lens 3, the second camera 4 and the second lens 5, the focusing unit includes the first motor 7, the first motion platform 6, the second The motor 9, the second moving platform 8 and the mirror 11, the computer and the control unit are PC.

Wherein the first motor 7 is connected with the first moving platform 6, the first motor 7 is used to drive the first moving platform 6 to move, the first lens 3 is installed on the first camera 2, and the first camera 2 is fixed on the first moving platform 6 on; the second lens 5 is installed on the second camera 4, and the second camera 4 is fixed on one side of the first moving platform 6; the second motor 9 is connected with the second moving platform 8, and the second motor 9 is used to drive the second The moving platform 8 moves, the object to be measured 10 is fixed on the second moving platform 8, the light source 1 is fixed on one side of the object to be measured 10, the reflector 11 is fixed on the other side of the object to be measured 10, and the reflector 11 is located on the second In the field of view of camera 4. The second motor 9 drives the movement direction of the second movement platform 8 to be parallel to the optical axis of the first camera 2 .

The stroke of the first motor 7 represents the stroke of the first moving platform 6, and the stroke of the second motor 9 represents the stroke of the second moving platform 8; the first lens 3 is used to collect the image of the object 10 to be measured, and the second lens 5 It is used to collect the image of the object 10 to be measured presented in the mirror 11 .

The real-time auto-focusing method of the camera under the vision system mainly includes two aspects of constructing a positive focus lookup table and real-time focusing.

For constructing the positive focus lookup table, the following steps are included:

S1. Build an offline fixed-focus system: in the offline state, in the vision system, fix the object 10 to be measured on the second motion platform 8, adjust the height of the light source 1 and the second motion platform 8, and ensure normal imaging;

S2. Determine the focusing range of the first camera 2: set the stroke of the first motor 7 to zero, drive the second motion platform 8 to reciprocate by the second motor 9, and find the positive focus A; then set the stroke of the first motor 7 to At the maximum value, the second motor 9 drives the second moving platform 8 to reciprocate to find the positive focus B; obtain the focusing range (A, B) of the first camera 2;

S3. Build a positive focus lookup table: divide the focus range (A, B) of the first camera 2 into equal parts according to the following formula:

M＝1+round((B-A)/step1)

Among them, round() is a rounding operation, and step1 is 1/3 of the depth of field of the first camera 2;

In this way, a positive focus lookup table Tab is created, and the positive focus lookup table Tab contains M elements, and the contents of each element include: the stroke of the first motor 7 and the position of the edge center when the focus is positive;

When determining the stroke of the first motor 7 and the position of the edge center when the focus is positive, it includes:

S3-1, drive the second moving platform 8 to position A; drive the stroke of the first motor 7 to move from the zero point to the maximum stroke position, and in this process, the first camera 2 and the second camera 4 collect pictures at the same time;

The image of the second camera 4 is binarized (using single threshold method and/or OSTU algorithm), then the edge of the object 10 to be tested is detected (using Canny operation and/or Sobel operation), and then the straight line y of the edge is fitted =k*x+b, finally calculate the center (x _c , y _c ) of the edge on the image, the calculation formula is as follows:

x _c = (Width-1)*0.5;

y _c =k*x _c +b;

Wherein, Width is the width of the image of the second camera 4;

Calculate the sharpness D of the image of the first camera 2, find the maximum sharpness D _max (that is, the positive focus), and record the edge center (x _c , y _c ) and the position of the first motor 7 in the positive focus lookup table Tab at this time. Travel position Pos;

S3-2, observe whether the second moving platform 8 has moved to the B end, if so, enter the next step, if not, drive the first moving platform 6 to step1 by the first motor 7, and then return to step S3-1;

S4. End. The construction of the positive focus lookup table Tab is completed.

After completing the construction of the positive focus lookup table Tab, the vision system can be used in actual production. In the online state, in the vision system, place the object 10 to be measured in the field of view of the first camera 2, power on and adjust the brightness of the light source 1 to make it image normally, and then start real-time focusing;

For real-time focusing, the following steps are included:

S5. Searching for a reference point: drive the first motor 7 to move from the zero point to the maximum travel position, and in the process, the first camera 2 and the second camera 4 collect pictures, and the images of the first camera 2 and the second camera 4 are in accordance with As described in S1, obtain the center (x _c , y _c ) of the edge of the positive focus, then traverse the positive focus lookup table Tab, find the point closest to it, and record it as the reference point R(x, y, Pos);

S6. Real-time focusing: collect pictures, calculate the center of the edge (x _c , y _c ), and calculate the absolute value of the difference between y _c and the y component of the reference point R d=abs(y _c -y), when d>setp 1, then traverse the positive focus lookup table Tab, find the point closest to the center (x _c , y _c ) of the current edge, obtain the stroke of the first motor 7 at this point, and finally drive the first motion platform 6 by the first motor 7 Then drive the first camera 2 to move to the stroke, and real-time focusing can be realized;

S7, when receiving the stop signal, enter the next step; otherwise, repeat the previous step;

S8. Exit the real-time focusing process.

The present invention has been described through the above-mentioned embodiments, but it should be understood that the above-mentioned embodiments are only for the purpose of illustration and description, and are not intended to limit the present invention to the scope of the described embodiments. For those skilled in the art, the present invention may have various modifications and changes, including combination and recombination of technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims (10)

1. A camera real-time automatic focusing method based on edge tracking is characterized in that: firstly, a vision system is built, and then a positive focus lookup table and real-time focusing are built;

the constructing of the positive focus lookup table comprises determining a focusing range of a camera in the vision system in an off-line state, and constructing the positive focus lookup table;

the real-time focusing comprises the steps that in an online state, in the visual system, an object to be measured is placed in the visual field of a camera, the camera is driven to move from a zero point to the maximum stroke, pictures are collected, the center position of the edge of a positive focus is obtained, a positive focus lookup table is traversed, and the point closest to the positive focus lookup table is determined as a reference point; then, acquiring a picture in real time, calculating an absolute value of a difference value between the center of the edge of the picture and a reference point, traversing the positive focus lookup table, searching a point closest to the center of the current picture edge, acquiring a camera stroke of the point, and driving the camera to the stroke, namely realizing real-time focusing.

2. The method for real-time auto-focusing of a camera based on edge tracking as claimed in claim 1, wherein: the vision system comprises an optical unit, an imaging unit and a focusing unit; the imaging unit includes a first camera and a second camera; the focusing unit comprises a first motion platform, a second motion platform and a reflector; the first camera is fixed on the first motion platform, and the second camera is fixed on one side of the first motion platform; the object to be detected is fixed on the second motion platform; the optical unit is fixed on one side of the object to be measured, the reflector is fixed on the other side of the object to be measured, and the reflector is positioned in the visual field of the second camera; the motion direction of the second motion platform is parallel to the optical axis of the first camera.

3. The method of claim 2 for real-time auto-focusing of an edge-tracking based camera, characterized by: in the process of building the positive focus look-up table,

first determining a focus range of a first camera, comprising: setting the stroke of the first motion platform to zero, driving the second motion platform to reciprocate and searching for a positive focus A; setting the stroke of the first motion platform to be maximum, driving the second motion platform to reciprocate, and searching for a positive focus B to obtain the focusing range (A, B) of the first camera;

then, constructing a positive focus lookup table, including: and equally dividing the focusing range of the first camera to construct a positive focus lookup table.

4. The method of claim 3 for real-time auto-focusing of an edge-tracking based camera, wherein: equally dividing the focusing range of the first camerase:Sub>A according to ase:Sub>A formulase:Sub>A M =1+ round ((B-A)/step 1);

wherein round () is a rounding operation, step1 is 1/5-1/2 of the depth of field of the first camera; the positive focus lookup table contains M elements, each element including a stroke and a position of an edge center of the first motion stage at positive focus.

5. The method for real-time auto-focusing of a camera based on edge tracking as claimed in claim 4, wherein: step1 is 1/3 of the depth of field of the first camera.

6. The method of claim 4 for real-time auto-focusing of an edge-tracking based camera, wherein:

in the building of the positive focus lookup table, determining the stroke and the position of the edge center of the first motion platform in the positive focus comprises:

firstly, driving a second motion platform to a position A, moving the first motion platform from a zero point to a maximum stroke, simultaneously acquiring pictures by a first camera and a second camera in the process, carrying out binarization on an image of the second camera, detecting the edge of an object to be detected, fitting a straight line of the edge, and finally calculating the center of the edge on the image;

then calculating the definition D of the image of the first camera, and searching the maximum definition D_maxAnd recording the current edge center and the stroke position of the first motion platform in the positive focus lookup table, observing whether the second motion platform moves to the point B, if so, finishing, and if not, continuously driving the second motion platform to move forwards until the second motion platform moves to the point B.

7. The method of claim 6, wherein the method comprises: the straight line of the edge is noted as y = k x + b, and finally the center of the edge on the image is calculated (x)_c,y_c) The calculation formula is as follows:

x_c＝(Width-1)*0.5；

y_c＝k*x_c+b；

wherein Width is the Width of the image of the second camera.

8. The method for real-time auto-focusing of a camera based on edge tracking as claimed in claim 6, wherein: methods of binarizing the image of the second camera include single threshold methods and/or the OSTU algorithm.

9. The method of claim 6, wherein the method comprises: the method for detecting the edge of the object to be detected comprises Canny operation and/or Sobel operation.

10. The method of claim 7 for real-time auto-focusing of an edge-tracking based camera, wherein: in the real-time focusing process,

first, a reference point is found: driving the first motion platform to move from the zero point to the maximum stroke, and acquiring pictures by the first camera and the second camera to acquire the center (x) of the edge of the positive focus_c,y_c) Then, traversing the positive focus lookup table, finding the point closest to the positive focus lookup table, and marking the point as a reference point R (x, y, pos);

and then, real-time focusing is carried out: the first camera takes a picture and calculates the center (x) of the edge_c,y_c) Calculating y_cThe absolute value of the difference from the y-component of the reference point R, d = abs (y)_cY), when d > ste1p, then traverse the positive focus lookup table, looking for the center (x) with the current edge_c,y_c) And (3) acquiring the stroke of the first motion platform of the nearest point, and then driving the first motion platform to drive the first camera to the stroke, so that real-time focusing can be realized.

Images