CN110376207B - An image acquisition method of an on-line detection system for surface defects of ultra-wide and thick plates - Google Patents

Abstract

The invention provides an image acquisition method of an online detection system for surface defects of an ultra-wide thick plate, which belongs to the technical field of machine vision nondestructive detection, wherein the number of linear array cameras used is determined according to the maximum width of a plate surface, the minimum size of defects to be detected and the camera resolution, the number of narrow band L ED light sources is consistent with the number of cameras, after a steel plate enters a camera acquisition area, a laser velocimeter acquires the speed of the steel plate so as to determine the acquisition linear velocity of the linear array cameras, then a multi-channel PWM signal generator is controlled by an industrial personal computer to generate PWM signals with the same pulse frequency and the acquisition linear velocity, the PWM signals with the same number as the cameras are output to control the cameras to acquire synchronously, and finally, images acquired by each camera are shifted and spliced in the length direction and the width direction to obtain a complete surface image.

Description

An image acquisition method of an on-line detection system for surface defects of ultra-wide and thick plates

technical field

The invention relates to the technical field of machine vision non-destructive testing, in particular to an image acquisition method of an on-line detection system for surface defects of ultra-wide and thick plates.

Background technique

At present, the online detection system of steel plate surface defects based on machine vision has been widely used in the production lines of major steel mills, which plays a key role in ensuring the surface quality of steel plates. The key of this system is to collect complete steel plate images with uniform background. for accurate defect identification. For the detection of ultra-wide boards larger than 5 meters, due to the requirements of detection accuracy, multiple cameras need to be arranged in the width direction to complete the acquisition of the entire board surface. These multiple cameras need to collect images in the width direction and the movement direction. Accurate splicing can facilitate subsequent defect analysis. Because the board surface is very wide, a strip of LED light source cannot guarantee the uniformity of the entire lighting area, which will cause the background brightness of the collected steel plate image to appear uneven in the width direction. Therefore, multiple strip-shaped LED light sources are needed to provide illumination. How to control the brightness changes of these LED light sources in time to ensure that the camera corresponding to the light source captures images with a uniform background needs to be considered.

SUMMARY OF THE INVENTION

The technical problem to be solved by the present invention is to provide an image acquisition method of an on-line detection system for super-wide and thick plate surface defects.

The method realizes image acquisition through a combination of a line scan camera, a narrow-band LED light source, a multi-channel PWM signal generator, a laser velocimeter, and a signal enhancement module. The implementation process includes the following steps:

(1) According to the maximum width L of the ultra-wide thick plate, the minimum size α of the defect to be detected, and the resolution M of the line scan camera, determine the number N of line scan cameras to be used, and the number of narrow-band LED light sources and the number of line scan cameras are maintained. consistent;

(2) After the ultra-wide and thick plate enters the collection area of the line scan camera, the laser tester obtains the speed ν of the ultra-wide and thick plate, and determines the acquisition line rate of the line scan camera according to the obtained speed ν of the ultra-wide and thick plate and the minimum size α of the defect to be detected f;

(3) Control the multi-channel PWM signal generator to generate PWM signals with the same pulse frequency as the acquisition line rate f, and output N pulse signals with the same number as the line scan cameras, and each signal is divided into two parts;

(4) By calibrating the position difference between each light band of the narrow-band LED light source in advance, the images collected by each line scan camera are spliced by displacement in the length direction;

(5) Shift and stitch the images collected by each line scan camera in the length and width directions to obtain a complete surface image.

Wherein, the number of line scan cameras in step (1) is N=2L/(α×M).

In step (2), the acquisition linear rate of the line scan camera is f=2ν/α. In order to ensure the detection of defects with a minimum size of α, the acquisition accuracy here needs to reach α/2.

In step (3), a part of each signal is sent to a line scan camera using an external trigger to achieve synchronous acquisition, and the other part is directly used as a power supply for a narrow-band LED light source after being enhanced by a signal enhancement module. As the power supply signal of the narrow-band LED light source, it can ensure that the brightness of the light source can be adjusted by controlling the PWM signal generator to change the duty cycle of the signal under the condition that the acquisition rate remains unchanged, that is, the acquisition accuracy remains unchanged.

The acquisition illumination areas of each group of line scan cameras are arranged in a staggered pattern.

Step (4) is specifically as follows: the dislocation distance of the illumination areas of two adjacent line scan cameras is σ, the collection accuracy θ in the length direction of the ultra-wide thick plate is α/2, and α is the minimum size of the defect to be detected, then the displacement is required. The number of pixels Δ is:

Δ=σ/θ=2σ/α.

The beneficial effects of the above-mentioned technical solutions of the present invention are as follows:

In the above scheme, multiple sets of line scan cameras and narrow-band LED light sources are used to obtain the accurate movement speed of the steel plate through a laser velocimeter, and PWM signals are used to realize the external trigger synchronous acquisition control of multiple cameras, so as to ensure the acquisition positions of multiple cameras, and simultaneously pass The pulse width of the same signal is used to control the brightness of the illumination light source of each line of image acquisition. This method can realize the adjustment of the light source brightness of each line of the acquired image. At the same time, the same PWM signal is used to control the acquisition speed and image brightness in hardware mode. , which can realize the brightness adjustment of a single line in the image, so that the light source can quickly respond to the request of brightness change.

Description of drawings

Fig. 1 is the overall scheme diagram of the image acquisition method of the ultra-wide and thick plate surface defect on-line detection system of the present invention;

FIG. 2 is a layout diagram of each camera acquisition area in an embodiment of the present invention;

FIG. 3 is a schematic diagram of a PWM signal in an embodiment of the present invention.

Detailed ways

In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, the following will be described in detail with reference to the accompanying drawings and specific embodiments.

The invention provides an image acquisition method of an on-line detection system for super-wide and thick plate surface defects.

As shown in Figure 1, the method realizes image acquisition through a combination of a line scan camera, a narrow-band LED light source, a multi-channel PWM signal generator, a laser velocimeter, and a signal enhancement module. The implementation process includes the following steps:

(1) According to the maximum width L of the ultra-wide and thick plate surface, the minimum size α of the defect to be detected, and the resolution M of the line scan camera, determine the number N of line scan cameras to be used, and the number of narrow-band LED light sources and the number of line scan cameras should be consistent ;

(2) After the ultra-wide and thick plate enters the collection area of the line scan camera, the laser tester obtains the speed ν of the ultra-wide and thick plate, and determines the acquisition line rate of the line scan camera according to the obtained speed ν of the ultra-wide and thick plate and the minimum size α of the defect to be detected f;

(3) Control the multi-channel PWM signal generator to generate PWM signals with the same pulse frequency as the acquisition line rate f, and output N pulse signals with the same number as the line scan cameras, and each signal is divided into two parts;

(4) By calibrating the position difference between each light band of the narrow-band LED light source in advance, the images collected by each line scan camera are spliced by displacement in the length direction;

(5) Shift and stitch the images collected by each line scan camera in the length and width directions to obtain a complete surface image.

The following description will be given in conjunction with specific embodiments.

According to the maximum width L of the board surface, the minimum size α of the defect to be detected, and the number of pixels M of the line scan camera used, determine the number of cameras to be used in the width direction N=2L/(αM);

At the same time, the number of narrow-band LED light sources is the same as the number of cameras.

The acquisition line rate f of the line scan camera is determined according to the obtained speed ν of the steel plate and the minimum size α in the width direction that needs to be detected for defects, f=2ν/α.

As shown in Figure 3, the acquired acquisition line rate f is used to control the multi-channel PWM signal generator to generate N PWM signals with the same pulse frequency as the acquisition line rate. Each signal is divided into two parts, and one part is sent to the external trigger In this way, the synchronous acquisition of N cameras can be ensured, and the consistent detection accuracy can be ensured in the width and length directions of the steel plate. The period value t of the signal is:

t=1/f

Another part of the PWM signal is amplified by the signal enhancement circuit and used as the power supply for the narrow-band LED light source matched with the camera. By calculating the average gray value of one or more lines collected by each camera in real time, the average gray value is evaluated and the benchmark is set. The difference in grayscale, sending an instruction to adjust the pulse width of the PWM signal, and adjusting the pulse width λ value of the PWM signal, it can be considered that there is a nearly linear relationship between the brightness ν of the light source and the pulse width λ.

ν=Kλ+β

K is a positive constant, and β is a constant.

Therefore, in the case of ensuring that the pulse frequency of the PWM signal remains unchanged, that is, to ensure that the acquisition rates of all cameras are the same, the pulse width value of the PWM signal that supplies power to the LED light source, that is, the duty cycle of the PWM signal can be changed. The brightness of the light source realizes the adjustment of the background brightness of the images collected by each camera. The entire adjustment process can be carried out by means of self-feedback to achieve the purpose of quickly adjusting the brightness of the light source.

In this way, the acquisition of all cameras and the illumination brightness of each camera are controlled through a set of PWM signals, so as to ensure the stitching accuracy of the images and the uniform background.

In order to avoid the mutual influence of the illumination light sources of each camera, the acquisition illumination area of each group of cameras adopts an interlaced pattern, as shown in Figure 2, which can ensure that the influence of mutual light sources can be avoided to a certain extent, because the entire acquisition process is precisely controlled according to the speed. The accuracy in the direction is obtained by calibrating the position difference between each light band in advance, and the images collected by each camera can be stitched by shifting in the length direction. For example, the dislocation distance of the illumination area of two adjacent cameras is The acquisition accuracy θ in the length direction is α/2, where α is the minimum size of defects to be detected, and the number of pixels to be shifted Δ is Δ=σ/θ=2σ/α.

In the specific application process, the workflow steps are as follows:

Step 1. The moving steel plate enters the collection area of the camera. (The camera adopts a line scan camera to achieve accurate stitching of images on the moving steel plate, and the collection area is very narrow, which can avoid the interference of different camera light sources). The laser velocimeter measures the result. The exact movement speed of the current steel plate.

Step 2. According to the design acquisition accuracy and the current speed of steel plate movement, calculate the acquisition line rate that the camera needs to use, and control the multi-channel PWM signal generator to generate multi-channel PWM pulse signals with the same frequency and the same number of cameras. It is sent to the line scan camera, and the other part is amplified by the signal power enhancement module to supply power to the LED light source. The camera adopts an external trigger method, and will start synchronous acquisition after receiving the signal.

Step 3: Calculate the average gray value of the single line or multi-line collected by each camera, and compare it with the set benchmark. If the difference value is large, adjust the pulse width of the PWM pulse signal of the channel, without changing the acquisition. The brightness of the light source is adjusted in time under the condition of the speed to ensure that the images with uniform background brightness of each camera are obtained.

Step 4: After the images of each camera are obtained, the images are shifted in the width direction according to the pre-calibrated overlapping positions, and the images of the entire surface are obtained by stitching after the shifting.

The above are the preferred embodiments of the present invention. It should be pointed out that for those skilled in the art, without departing from the principles of the present invention, several improvements and modifications can be made. These improvements and modifications It should also be regarded as the protection scope of the present invention.

Claims (6)

1. an ultra-wide thick plate surface defect on-line detection system image acquisition method, it is characterized in that: this method realizes ultra-high by the combination of line scan camera and narrow-band LED light source, multi-channel PWM signal generator, laser velocimeter, signal enhancement module The acquisition process of wide and thick plate images includes the following steps: (1) According to the maximum width L of the ultra-wide thick plate, the minimum size α of the defect to be detected, and the resolution M of the line scan camera, determine the number N of line scan cameras used, and the number of narrow-band LED light sources and the number of line scan cameras are maintained. consistent; (2) After the ultra-wide and thick plate enters the collection area of the line scan camera, the laser velocimeter obtains the speed ν of the ultra-wide and thick plate, and the acquisition line rate of the line scan camera is determined according to the obtained speed ν of the ultra-wide and thick plate and the minimum size α of the defect to be detected. f; (3) The industrial computer controls the multi-channel PWM signal generator to generate PWM signals with the same pulse frequency as the acquisition line rate f, and output N pulse signals with the same number as the line scan cameras, and each signal is divided into two parts; (4) By calibrating the position difference between each light band of the narrow-band LED light source in advance, the images collected by each line scan camera are spliced by displacement in the length direction; (5) Shift and stitch the images collected by each line scan camera in the length and width directions to obtain a complete surface image; In the step (3), a part of each signal is sent to a line scan camera using an external trigger to realize synchronous acquisition, and the other part is directly used as a power supply for a narrow-band LED light source after being enhanced by a signal enhancement module. 2 . The image acquisition method of an on-line detection system for super-wide and thick plate surface defects according to claim 1 , wherein the number of line scan cameras in the step (1) is N=2L/(α×M). 3 . 3 . The image acquisition method of an on-line detection system for super-wide and thick plate surface defects according to claim 1 , characterized in that: in the step (2), the acquisition line rate of the line scan camera is f=2ν/α. 4 . 4 . The image acquisition method of an on-line detection system for super-wide and thick plate surface defects according to claim 1 , wherein the acquisition and illumination areas of each line scan camera are arranged in a staggered pattern. 5 . 5 . The image acquisition method of an on-line detection system for super-wide and thick plate surface defects according to claim 1 , wherein the step (4) is specifically as follows: the dislocation distance of the illumination areas of two adjacent line scan cameras is σ , which is σ . 6 . To ensure that defects with a minimum size of α can be detected, and the acquisition accuracy θ in the length direction of the ultra-wide thick plate is α/2, the number of pixels to be shifted Δ is: Δ=σ/θ=2σ/α. 6. The method for image acquisition of an on-line detection system for ultra-wide and thick plate surface defects according to claim 1, characterized in that: the enhanced PWM signal powered by the narrow-band LED light source can ensure that when the acquisition rate is unchanged, that is, the acquisition accuracy is unchanged, By controlling the PWM signal generator to change the signal duty cycle, the brightness of the light source can be adjusted so as to quickly adjust the brightness of the image captured by the camera.

Images