Machine vision light source adaptive dynamic dimming method and device and computer equipment

Abstract

The invention discloses a machine vision light source self-adaptive dynamic dimming method and device and computer equipment, and the method comprises the steps: collecting an image sequence, constructing a multivariate linear regression equation of three variables, namely illumination intensity, light source color and illumination direction, related to image features, obtaining a light source parameter corresponding to the maximum value of the image features through fitting calculation, and carrying out the self-adaptive dynamic dimming of a machine vision light source. The light source is controlled to carry out self-adaptive dynamic dimming, so that the target object is under the optimal illumination intensity, it is guaranteed that the collected surface picture of the target object has the better foreground and background contrast ratio, the optimal sharpness and the least noise, and the image quality is effectively guaranteed.

Description

Technical Field

[0001] The present invention belongs to the field of machine vision technology, and in particular relates to a method, device and computer equipment for adaptive dynamic dimming of a machine vision light source. Background Art

[0002] Machine vision technology aims to replace the human eye in defect detection, dimension measurement, target positioning and other tasks through image capture devices and computer vision algorithms. With the development of machine vision technology, more and more manufacturing companies are applying machine vision technology to assembly lines to improve factory automation and intelligence, and reduce labor costs and missed detection rates.

[0003] As a pre-processing device for image acquisition, the light source plays a vital role in the image quality. However, the incoming materials of assembly line production are often highly random. For the surfaces of objects with different colors, roughness and materials, the same light intensity, light color or light source type often cannot provide sufficient light, resulting in reduced detection accuracy. In addition, too high or too low light intensity will have a negative impact on image quality, increase image noise or cause object details to be lost, making it difficult for subsequent visual work to achieve the best results. There are several major problems with existing machine vision systems: 1. The single light source intensity setting cannot adapt to the diverse detection environment and object characteristics, and it is difficult to maintain detection stability in complex scenes; 2. The traditional light source intensity adjustment method lacks a dynamic adjustment mechanism and cannot be optimized immediately according to the changes in the real-time detection object, resulting in slow system response and poor flexibility; 3. The method of manually adjusting the intensity according to the incoming material type is inefficient, affected by subjective factors, has high requirements on the professional ability of the operator, and it is difficult to ensure consistency, which increases the unreliability and operating cost of the system. Summary of the invention

[0004] To solve the above problems, the purpose of the present invention is to provide a method, device and computer equipment for adaptive dynamic dimming of machine vision light sources, which can realize image feature acquisition, data curve fitting, optimal lighting matching by collecting image sequences, and adjust the light source combination and the brightness, color and direction of the light source, so as to effectively ensure image quality.

[0005] In order to achieve the above-mentioned invention object, the present invention adopts the following technical scheme:

[0006] A method for adaptive dynamic dimming of a machine vision light source comprises the following steps:

[0007] S1. Define light source parameter variables: X is light intensity, Y is light color, and Z is light angle. Combining variables Y and Z can obtain several groups of different optical properties.

[0008] S2, modify the variable X, use K equally divided sampling points, including J1, J2, ..., Jp, and combine them with the several groups of optical properties in step S1, and control the image acquisition device to obtain Q image sequences I(i) in the same area on the surface of the target object, i = 0, 1, 2, ..., Q-1, where the coordinates of i in the independent variable system are described as (X, Y, Z);

[0009] S3, select a group of image sequences I(j) under the same X value, the total number of j is the number of images captured in the sequence, and the image index is 1 to the total number; calculate the image features of the image sequence I(j) respectively, where the evaluation index of the image features is:

[0010] (1) Define the sharpness parameter F of the image, and use the convolution kernels in the X direction, Y direction, 45° oblique direction, and 135° oblique direction to perform convolution operations on the image, and obtain the convolution kernel results as Gx, Gy, G 45 and G 135 , define the result G of the convolution center,

[0011] G=|Gx|+|Gy|+|G 45 |+|G 135 |

[0012] Convolution operation formula:

[0013]

[0014] Where i is the x-coordinate of a two-dimensional image point; j is the y-coordinate of a two-dimensional image point; M is the convolution kernel; I(x, y) is the pixel value of the image at (x, y); M(x, y) is the element of the convolution kernel at (x, y);

[0015] After the convolution result of a single full image is calculated, the threshold T with the largest threshold classification variance is calculated according to the maximum inter-class variance method, and then the sharpness parameter F of the image is calculated.

[0016]

[0017] (2) Define the image contrast parameter C and calculate the expected value μ of a single image.

[0018]

[0019] Where n is the total number of pixels in the image; I(c, y) is the pixel value of the image at (x, y);

[0020] Calculate the image standard deviation σ,

[0021]

[0022] Define contrast parameter C = σ;

[0023] (3) The evaluation index of image features is S = a × F + b × C, where a and b are parameter weight coefficients, and a + b = 1. The weight classification ratio is selected according to the subsequent task type;

[0024] S4. Calculate the image sequence I(j) to obtain the image feature evaluation index result sequence when the variable X is one of the K equally divided sampling points, and obtain the maximum value S max , the index of variables Y and Z is (Y best , Z best )=argmax(S max ), is the optimal parameter value of Y and Z variables; Ji is the i-th sampling point; in (Y best , Z best ) state, select the image sequence in (Y best , Z best ) state, containing multiple images corresponding to K equally divided sampling points of variable X, with coordinates described as (X1, Y best , Z best )、(X2,Y best , Z best ),……,(Xk,Y best , Z best ), respectively calculate the evaluation index S of the image features, construct the coordinate system of X-S, where X is the horizontal axis independent variable, the calculated image evaluation index S is the vertical axis dependent variable, obtain the multi-point data of K equally divided sampling points corresponding to the variable X in the coordinate system, use a high-order polynomial fitting curve to obtain the X value where S obtains the maximum value within the range of light intensity X, and obtain the optimal X parameter value;

[0025] S5. Obtain the optimal optical parameters (X, Y, Z) for the surface of the target object, apply them to subsequent visual tasks, and build a corresponding database according to the properties of the target.

[0026] Furthermore, the grayscale value range of the above-mentioned light intensity X is 0-255; the range of the light color Y is [R, G, B, W], where R, G, B, W represent red, green, blue, and white light respectively; the light angle Z is divided into several angles, and the range of the light angle Z is [θ1, θ2, ..., θ n ]; According to the selection of the variable Y and the variable Z, multiple groups of different optical properties are obtained.

[0027] A machine vision light source adaptive dynamic dimming device comprises a light source, a light source control device, and an image acquisition device, wherein:

[0028] The light source includes a plurality of LED lamp beads for providing light sources, each of which has three different color light transmitters, red, green and blue, and is controlled in different areas; the optimal image quality is obtained by adjusting three variables, namely, input current, light source color signal and lamp bead direction signal;

[0029] The light source control device comprises a controller, a power module, a signal input module and an output module. The controller is used to control the on / off combination mode of multiple lamp beads of the light source and the color channel switching of the lamp beads; the power module is used to control the current and voltage of the light source to change the light intensity of the light source; the signal input module is used to receive external control signals; and the signal output module is used to send control signals to the controller.

[0030] The image acquisition device is used to acquire the surface image of the target object; at the same time, the light is modulated in cooperation with the industrial lens so that a higher quality image is formed on the image acquisition device.

[0031] Furthermore, the above-mentioned image acquisition device includes but is not limited to photosensitive elements such as CCD cameras and CMOS cameras.

[0032] Furthermore, the above-mentioned industrial lens includes but is not limited to FA lens, industrial telecentric lens, industrial macro lens and the like, which are devices used to change the light beam.

[0033] The present invention also provides a computer device, which includes a processor and a memory, wherein the memory is used to store instructions, and the processor communicates with the memory through a bus to execute the instructions stored in the memory to execute the instruction steps of the above-mentioned machine vision light source adaptive dynamic dimming method.

[0034] Due to the adoption of the above-mentioned technical solution, the present invention has the following advantages:

[0035] The method for adaptive dynamic dimming of machine vision light sources collects image sequences, constructs a multivariate linear regression equation of three variables of light intensity, light source color, and light direction for image features, obtains light source parameters corresponding to the maximum value of image features through fitting calculation, and controls the light source for adaptive dynamic dimming, so that the target object is under the optimal light intensity, ensuring that the collected surface image of the target object has better foreground-background contrast, optimal sharpness, and minimum noise, thereby improving the accuracy of subsequent visual tasks, reducing labor costs, improving the automation and intelligence of production lines, and greatly improving the adaptability of machine vision in various scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 is a flow chart of the method for adaptive dynamic dimming of a machine vision light source of the present invention;

[0037] Figure 2It is a structural diagram of the adaptive dynamic dimming device of the machine vision light source of the present invention;

[0038] Figure 3 1 is a schematic diagram of image sequence combination in an embodiment of a method for adaptive dynamic dimming of a machine vision light source of the present invention;

[0039] Figure 4 It is a structural diagram of a light source of a machine vision light source adaptive dynamic dimming device of the present invention;

[0040] In the picture: 1-red lamp bead; 2-green lamp bead; 3-blue lamp bead. DETAILED DESCRIPTION

[0041] The technical solution of the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments.

[0042] A method for adaptive dynamic dimming of a machine vision light source comprises the following steps:

[0043] S1. Define light source parameter variables: X is light intensity, Y is light color, and Z is light angle. Variables Y and Z are combined to obtain several groups of different optical properties. The grayscale value range of the light intensity X is 0 to 255. The range of the light color Y is [R, G, B, W], where R, G, B, and W represent red, green, blue, and white light, respectively. The light angle Z is divided into several angles.

[0044] S2, modify the variable X, use K equally divided sampling points, including J1, J2, ..., Jp, and combine them with the several groups of optical properties in step S1, control the image acquisition device to obtain Q image sequences I(i) in the same area on the surface of the target object, i = 0, 1, 2, ..., Q-1, wherein the coordinates of i in the independent variable system are described as (X, Y, Z); preferably, they are arranged according to the independent variable system coordinates (X, Y, Z), with X as the first coordinate, Y as the second coordinate, and Z as the third coordinate; the more equally divided points the variable X has, the higher the accuracy and the better the effect, but the amount of calculation increases, therefore, select appropriate equally divided sampling points, preferably, five equal divisions, six equal divisions, or seven equal divisions;

[0045] S3. Select a group of image sequences I(j) under the same X value, where the total number of j is the number of images captured in the sequence, and the image index is 1 to the total number, i.e. j=1 is the first image, j=2 is the second image, and so on; calculate the image features of the image sequence I(j) respectively, where the evaluation index of the image features is:

[0046] (1) Define the sharpness parameter F of the image, and use the convolution kernels in the X direction, Y direction, 45° oblique direction, and 135° oblique direction to perform convolution operations on the image, and obtain the convolution kernel results as Gx, Gy, G 45 and G135 , define the result G of the convolution center,

[0047] G=|Gx|+|Gy|+|G 45 |+|G 135 |

[0048] Convolution operation formula:

[0049]

[0050] Where, j is the x-coordinate of the coordinate point of the two-dimensional image; j is the y-coordinate of the coordinate point of the two-dimensional image; M is the convolution kernel; I(x, y) is the pixel value of the image at (x, y); M(x, y) is the element of the convolution kernel at (x, y);

[0051] After the convolution result of a single full image is calculated, the threshold T with the largest threshold classification variance is calculated according to the maximum inter-class variance method, and then the sharpness parameter F of the image is calculated.

[0052]

[0053] (2) Define the image contrast parameter C and calculate the expected value μ of a single image.

[0054]

[0055] Where n is the total number of pixels in the image; I(x, y) is the pixel value of the image at (x, y);

[0056] Calculate the image standard deviation σ,

[0057]

[0058] Define contrast parameter C = σ;

[0059] (3) Evaluation index of image features S = a × F + b × C, where a and b are parameter weight coefficients, and a + b = 1. The weight classification ratio is selected according to the subsequent task type; preferably, for defect detection tasks, b = 0.8, a = 0.2; for size measurement tasks, a = 0.8, b = 0.2;

[0060] S4. Calculate the image sequence I(j) to obtain the image feature evaluation index result sequence when X is one of the K equally divided sampling points, and obtain the maximum value S max , the index of variables Y and Z is (Y best , Z best )=argmax(S max ), is the optimal parameter value of Y and Z variables, and is the i-th sampling point; in (Y best , Z best) state, select the image sequence in (Y best , Z best ) state, containing multiple images corresponding to K equally divided sampling points of variable X, with coordinates described as (X1, Y best , Z best )、(X2,Y best , Z best )、......、(Xk,Y best , Z best ), respectively calculate the evaluation index S of the image features, construct a coordinate system of XS, wherein X is the horizontal axis independent variable, the calculated image evaluation index S is the vertical axis dependent variable, obtain the multi-point data of K equally divided sampling points corresponding to the variable X in the coordinate system, adopt a high-order polynomial fitting curve, obtain the X value at which S obtains the maximum value within the range of light intensity X, and obtain the optimal X parameter value; preferably, under the balance of fitting accuracy and calculation speed, multiple polynomial fitting is adopted, because the quadratic accuracy is low and the calculation speed of more than three times is slow, therefore, the cubic polynomial fitting is preferred;

[0061] S5. Obtain the optimal optical parameters (X, Y, Z) for the surface of the target object, apply them to subsequent visual tasks, and build a corresponding database according to the properties of the target; preferably, the target properties include but are not limited to color, roughness, reflectivity, and material, which can be directly indexed to its optimal optical parameters, and the corresponding optimal optical parameters are built according to the properties of the target surface.

[0062] The steps of the method for adaptive dynamic dimming of a machine vision light source of the present invention are described in detail below through an embodiment.

[0063] like Figure 1 , 2 As shown, a method for adaptive dynamic dimming of a machine vision light source includes the following specific steps:

[0064] S1. Define light source parameter variables: X is the grayscale value range of light intensity, which is 0-255; Y is the light color, with a range of [R, G, B, W], representing red, green, blue, and white light respectively; Z is the light angle, which is divided into five angles, with a range of [20°, 38°, 55°, 73°, 90°]. According to the selection of the above variables Y and Z, there are four types of Y and five types of Z. The permutations and combinations result in a total of 20 groups of different optical property formulas, that is, in this embodiment, Including (R, 20°), (R, 38°), (R, 55°), (R, 73°), (R, 90°), (G, 20°), (G, 38°), (G, 55°), (G, 73°), (G, 90°) , (B, 20°), (B, 38°), (B, 55°), (B, 73°), (B, 90°), (W, 20°), (W, 38°), (W, 55°), (W, 73°), (W, 90°);

[0065] S2, modify the variable X, use five equally divided sampling points, including 51, 102, 153, 204, 255, and combine them with the 20 sets of optical formulas in step S1, and control the image acquisition device to obtain 100 image sequences I(i) in the same area on the surface of the target object, i = 0, 1, 2, ..., 99, where the coordinates of i in the independent variable system are described as (X, Y, Z);

[0066] S3. Select a group of image sequences under the same X value, select the image sequence I(j) under X=102, j=20, 21, 22, ..., 39, and calculate the image features under the image sequence respectively, where the evaluation index of the image feature is:

[0067] (1) Define the sharpness parameter F of the image, and use the convolution kernels in the X direction, Y direction, oblique 45° and oblique 135° to perform convolution operations on the image. The convolution kernel types are:

[0068]

[0069] The convolution kernel results are Gx, Gy, G respectively. 45 and G 135 , define the result G of the convolution center,

[0070] G=|Gx|+|Gy|+|G 45 |+|G 135 |

[0071] Convolution operation formula:

[0072]

[0073] Where, j is the x-coordinate of the coordinate point of the two-dimensional image; j is the y-coordinate of the coordinate point of the two-dimensional image; M is the convolution kernel; I(x, y) is the pixel value of the image at (x, y); M(x, y) is the element of the convolution kernel at (x, y);

[0074] After the convolution result of a single full image is calculated, the threshold T with the largest threshold classification variance is calculated according to the maximum inter-class variance method, and then the sharpness parameter F of the image is calculated.

[0075]

[0076] (2) Define the image contrast parameter C and calculate the expected value μ of a single image.

[0077]

[0078] Where n is the total number of pixels in the image; I(x, y) is the pixel value of the image at (x, y);

[0079] Calculate the image standard deviation σ,

[0080]

[0081] Define contrast parameter C = σ;

[0082] (3) The evaluation index of image features is S = a × F + b × C, where a and b are parameter weight coefficients, and a + b = 1. The weight classification ratio is selected according to the subsequent task type;

[0083] S4. Calculate the image sequence I(j) to obtain the image feature evaluation index result sequence when X=102, and obtain the maximum value S max , the index of variables Y and Z is (Y best , Z best )=argmax(S max ), is the optimal parameter value of Y and Z variables; in (Y best , Z best ) state, select the image sequence in (Y best , Z best ) state, contains 5 images, the coordinates are described as (51, Y best , Z best )、(102,Y best , Z best )、(153,Y best , Z best )、(204,Y best , Z best )、(255,Y best , Z best), respectively calculate the evaluation index S of the image features, construct the X-S coordinate system, where X is the horizontal axis independent variable, the calculated image evaluation index S is the vertical axis dependent variable, obtain the five-point data in the coordinate system, use a high-order polynomial fitting curve to obtain the X value where S obtains the maximum value when the illumination intensity X is in the range of 0 to 255 grayscale, and obtain the optimal X parameter value;

[0084] S5. Obtain the optimal optical parameters (X, Y, Z) for the surface of the target object, apply them to subsequent visual tasks, and build a corresponding database according to the properties of the target; control the light source through the light source controller to illuminate with the optimal light source parameters to achieve adaptive dynamic dimming, so that the workpiece is at the optimal light intensity.

[0085] like Figure 3 As shown, a machine vision light source adaptive dynamic dimming device includes a light source, a light source control device, and an image acquisition device, wherein:

[0086] The light source includes LED lamp beads for providing light sources, each of which has three different color light transmitters: red, green and blue, and is controlled in different areas, namely, red lamp bead 1, green lamp bead 2, and blue lamp bead 3. The LED lamp beads are arranged at 20° to 90°, such as Figure 4 As shown in the figure, the best image quality can be obtained by adjusting the three variables of input current, light source color signal and lamp bead direction signal.

[0087] The light source control device comprises a controller, a power module, a signal input module and an output module. The controller is used to control the on / off combination mode of multiple lamp beads of the light source and the color channel switching of the lamp beads; the power module is used to control the current and voltage of the light source to change the light intensity of the light source; the signal input module is used to receive external control signals; and the signal output module is used to send control signals to the controller.

[0088] The image acquisition device includes but is not limited to photosensitive elements such as CCD cameras and CMOS cameras, which are used to obtain the surface image of the target object; at the same time, the light is modulated in conjunction with an industrial lens so that a higher quality image is formed on the image acquisition device. The industrial lens includes but is not limited to FA lenses, industrial telecentric lenses, industrial macro lenses and other devices for changing the light beam.

[0089] The present invention also provides a computer device, which includes a processor and a memory, wherein the memory is used to store instructions. The processor communicates with the memory through a bus and is used to execute the instructions stored in the memory to execute the instruction steps of the above-mentioned machine vision light source adaptive dynamic dimming method, and connect the bus to the light source control device of the second aspect to transmit the dimming signal to the light source control device to realize the dimming process.

[0090] The above description is only a preferred embodiment of the present invention, and is not intended to limit the present invention. Without departing from the spirit and scope of the present invention, all equivalent changes and modifications made within the scope of the patent application of the present invention shall fall within the scope of patent protection of the present invention.

Claims

1. A method for adaptive dynamic dimming of a machine vision light source, characterized by: It includes the following steps: S1. Define light source parameter variables: X is light intensity, Y is light color, and Z is light angle. Combining variables Y and Z can obtain several groups of different optical properties. S2, modify the variable X, use K equally divided sampling points, including J1, J2, ..., Jp, and combine them with the several groups of optical properties in step S1, and control the image acquisition device to obtain Q image sequences I(i) in the same area on the surface of the target object, i = 0, 1, 2, ..., Q-1, where the coordinates of i in the independent variable system are described as (X, Y, Z); S3, select a group of image sequences I(j) under the same X value, the total number of j is the number of images captured in the sequence, and the image index is 1 to the total number; calculate the image features of the image sequence I(j) respectively, where the evaluation index of the image features is: (1) Define the sharpness parameter F of the image, and use the convolution kernels in the X direction, Y direction, 45° oblique direction, and 135° oblique direction to perform convolution operations on the image, and obtain the convolution kernel results as Gx, Gy, G 45 and G 135 , define the result G of the convolution center, G=|Gx|+|Gy|+|G 45 |+|G 135 | Convolution operation formula: Where, j is the x-coordinate of the coordinate point of the two-dimensional image; j is the y-coordinate of the coordinate point of the two-dimensional image; M is the convolution kernel; I(x,y) is the pixel value of the image at (x,y); M(x,y) is the element of the convolution kernel at (x,y); After the convolution result of a single full image is calculated, the threshold T with the largest threshold classification variance is calculated according to the maximum inter-class variance method, and then the sharpness parameter F of the image is calculated. (2) Define the image contrast parameter C and calculate the expected value μ of a single image. Where n is the total number of pixels in the image; I(x,y) is the pixel value of the image at (x,y); Calculate the image standard deviation σ, Define contrast parameter C = σ; (3) The evaluation index of image features is S = a × F + b × C, where a and b are parameter weight coefficients, and a + b = 1. The weight classification ratio is selected according to the subsequent task type; S4. Calculate the image sequence I(j) to obtain the image feature evaluation index result sequence when the variable X is one of the K equally divided sampling points, and obtain the maximum value S max , the index of variables Y and Z is (Y best ,Z best )=argmax(S max ), is the optimal parameter value of Y and Z variables; Ji is the i-th sampling point; in (Y best ,Z best ) state, select the image sequence in (Y best ,Z best ) state, including multiple images corresponding to K equally divided sampling points of variable X, with coordinates described as (X1,Y best ,Z best )、(X2,Y best ,Z best ),……,(Xk,Y best ,Z best ), respectively calculate the evaluation index S of the image features, construct the coordinate system of X-S, where X is the horizontal axis independent variable, the calculated image evaluation index S is the vertical axis dependent variable, obtain the multi-point data of K equally divided sampling points corresponding to the variable X in the coordinate system, use a high-order polynomial fitting curve to obtain the X value where S obtains the maximum value within the range of light intensity X, and obtain the optimal X parameter value; S5. Obtain the optimal optical parameters (X, Y, Z) for the surface of the target object, apply them to subsequent visual tasks, and build a corresponding database according to the properties of the target.

2. The method for adaptive dynamic dimming of a machine vision light source according to claim 1, characterized in that: The grayscale value range of the light intensity X is 0-255; the range of the light color Y is [R, G, B, W], where R, G, B, W represent red, green, blue, and white light respectively; the light angle Z is divided into several angles, and the range of the light angle Z is [θ1, θ2, ..., θ n ]; According to the selection of the variable Y and the variable Z, multiple groups of different optical properties are obtained.

3. A machine vision light source adaptive dynamic dimming device, used to implement the machine vision light source adaptive dynamic dimming method according to claim 1 or 2, characterized in that: It includes a light source, a light source control device, and an image acquisition device, wherein: The light source includes a plurality of LED lamp beads for providing light sources, each of which has three different color light transmitters, red, green and blue, and is controlled in different areas; the optimal image quality is obtained by adjusting three variables, namely, input current, light source color signal and lamp bead direction signal; The light source control device comprises a controller, a power module, a signal input module and an output module. The controller is used to control the on / off combination mode of multiple lamp beads of the light source and the color channel switching of the lamp beads; the power module is used to control the current and voltage of the light source to change the light intensity of the light source; the signal input module is used to receive external control signals; and the signal output module is used to send control signals to the controller. The image acquisition device is used to acquire the surface image of the target object; at the same time, the light is modulated in cooperation with the industrial lens so that a higher quality image is formed on the image acquisition device.

4. The machine vision light source adaptive dynamic dimming device according to claim 3, characterized in that: The image acquisition device includes but is not limited to photosensitive elements such as CCD cameras and CMOS cameras.

5. The machine vision light source adaptive dynamic dimming device according to claim 3, characterized in that: The industrial lens includes but is not limited to FA lens, industrial telecentric lens, industrial macro lens and other devices used to change the light beam.

6. A computer device, characterized in that: It includes a processor and a memory, the memory is used to store instructions, the processor communicates with the memory through a bus, and is used to execute the instructions stored in the memory to execute the instruction steps of the machine vision light source adaptive dynamic dimming method as described in claim 1 or 2.

Images