A multi-dimensional identification method and system for rolled parts at the outlet of a water tank

Abstract

This invention discloses a multi-dimensional identification method for rolled pieces at the outlet of a water tank, relating to the identification technology of double-high wire rod rolled pieces in steel production. The method involves acquiring horizontal and vertical images; performing edge detection and image combination processing on these two images to obtain multi-dimensional image combinations; selecting the image combination with the largest area from all multi-dimensional image combinations as the baseline image combination; performing similarity analysis between the baseline image combination and other multi-dimensional image combinations to extract approximate image combinations; if the number of approximate image combinations is less than or equal to a counting threshold, activating the warning system for the corresponding water tank. This invention also discloses a multi-dimensional identification system for rolled pieces at the outlet of a water tank. This invention avoids the problem of black edges on the rolled pieces interfering with the detection signal, ensures the synchronization of rolled pieces at the outlets of each water tank, and solves the problem in existing technologies where the detection of rolled pieces at the water tank outlet via sensing components easily leads to signal interruption.

Description

Technical Field

[0001] This invention relates to the identification technology of double-high wire rod rolled pieces in steel production, and more specifically, to a multi-dimensional identification method and system for rolled pieces at the outlet of a water tank. Background Technology

[0002] The pre-cooling system (i.e., water tank number zero, hereinafter referred to as the water tank) in the double-high-speed rolling mill is crucial for cooling the rolled pieces. Before entering the water tank, the rolled pieces have only undergone rolling on the four pre-finishing mill stands and have not yet undergone the finishing rolling process, resulting in a relatively large size. Due to the large cross-sectional area of ​​the rolled pieces, the water flow in the water tank is kept constant, and the water volume is adjusted to its maximum. However, this leads to a serious bottleneck problem: due to the positional changes of the rolled pieces and the continuous, maximum flow of water in the water tank, irregular black edges appear on the surface of the rolled pieces exiting the water tank. The location of these irregular black edges is not fixed, which significantly impacts the inspection of the rolled pieces at the water tank outlet. The main problem is that existing technology for rolling mill detection involves using a sensor to detect the high-temperature luminous edge of the rolling mill at a fixed position. When the detection medium of the sensor comes into contact with the irregular black edge of the rolling mill, the detection medium often fails to be transmitted to the sensor, resulting in an interruption of the rolling mill detection signal. Consequently, the rolling mills in each water tank become out of sync, causing subsequent rolling mill tracking and related interlocking controls to malfunction, leading to steel accumulation in the high-speed zone and affecting normal production operations. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a multi-dimensional identification method and system for rolled pieces at the outlet of a water tank, which addresses the shortcomings of the prior art. This solves the problem that the detection signal is easily interrupted and the synchronization of the rolled pieces is affected by the use of sensing components to detect rolled pieces at the outlet of a water tank in the prior art.

[0004] The multi-dimensional identification method for rolled pieces at the outlet of a water tank according to the present invention includes the following steps:

[0005] Step 1: When any sensing component is triggered, acquire horizontal and vertical images of the rolled pieces at all water tank outlets along the double-height line.

[0006] Step 2: Perform edge detection and image combination processing on the horizontal and vertical images to obtain a multi-dimensional image combination;

[0007] Step 3: Select the image with the largest horizontal area from all the multi-dimensional image combinations as the reference image combination;

[0008] Step 4: Perform similarity analysis on the baseline image combination with other multi-dimensional image combinations to extract approximate image combinations from the multi-dimensional image combinations;

[0009] Step 5: If the number of approximate image combinations is less than or equal to a set counting threshold, activate the early warning system of the water tank corresponding to the approximate image combination;

[0010] If the number of approximate image combinations exceeds the set counting threshold, then the image with the second largest horizontal area in the multi-dimensional image combination is taken as the benchmark image combination, and the process returns to step four.

[0011] As a further improvement, if the number of approximate image combinations exceeds the set counting threshold after using the second largest multi-dimensional image combination as the reference image combination in steps four and five, a fault alarm will be triggered, and the conveying of the rolled piece at the water tank outlet will be stopped.

[0012] As a further improvement, the specific detection method for the edge detection is as follows:

[0013] The first step is to perform Gaussian filtering on the horizontal and vertical images;

[0014] The second step is to use edge detection operators to highlight the local edges of the horizontal and vertical images obtained in the first step.

[0015] The third step is to define the edge intensity of the pixels and extract the edge point sets of the horizontal and vertical images obtained in the second step by threshold segmentation.

[0016] Step 4: Based on the edge point set, determine the final basic unit edge position according to the distribution of adjacent edge pixels, and draw the edge path image according to the basic unit edge position;

[0017] Step 5: Extract the outermost edge path from the edge path image as the target edge, and remove other edge paths within the target edge to obtain the horizontal target image and the vertical target image.

[0018] Furthermore, the edge detection operator is the Canny operator.

[0019] Furthermore, the specific combination method of the image combination is as follows:

[0020] The horizontal target image is incorporated into the vertical target image, and the target edges in the two target images are made independent of each other.

[0021] Furthermore, the image area is the area enclosed by all target edges in the multi-dimensional image combination.

[0022] As a further improvement, the similarity analysis specifically involves:

[0023] A similarity threshold is set. If the contour similarity between the baseline image combination and the multi-dimensional image combination is greater than the similarity threshold, then the multi-dimensional image combination is similar to the baseline image combination; otherwise, the multi-dimensional image combination is regarded as an approximate image combination.

[0024] A multi-dimensional identification system for rolled pieces at the outlet of a water tank includes a horizontal image acquisition device, a vertical image acquisition device, and a controller; the controller is electrically connected to both the horizontal and vertical image acquisition devices, and the controller processes the images acquired by the horizontal and vertical image acquisition devices using the multi-dimensional identification method described above.

[0025] Beneficial effects

[0026] The advantages of this invention are as follows: by performing edge detection and image combination processing on the acquired horizontal and vertical images, similarity analysis is performed on the resulting multi-dimensional image combination, and counting threshold analysis is performed on the extracted approximate image combination. This effectively avoids the problem of black edges on the rolled piece interfering with the detection signal, ensures the synchronization of the rolled pieces at the outlet of each water tank, and solves the problem that the detection signal is easily interrupted when the rolled piece at the outlet of the water tank is detected by the sensing component in the prior art. Attached Figure Description

[0027] Figure 1 This is a flowchart of the multi-dimensional recognition method of the present invention;

[0028] Figure 2 This is a flowchart of the edge detection process of the present invention;

[0029] Figure 3 This is a flowchart of the edge detection process based on the Canny operator of the present invention. Detailed Implementation

[0030] The present invention will be further described below with reference to embodiments, but this does not constitute any limitation on the present invention. Any limited modifications made by any person within the scope of the claims of the present invention are still within the scope of the claims of the present invention.

[0031] See Figure 1 The present invention provides a multi-dimensional identification method for rolled parts at the outlet of a water tank, comprising the following steps.

[0032] Step 1: When any sensing component is triggered, acquire horizontal and vertical images of the rolled pieces at all water tank outlets along the double-height line. The horizontal image is the top surface image of the rolled piece, and the vertical image is the side image of the rolled piece.

[0033] Step 2: Perform edge detection and image combination processing on the horizontal and vertical images to obtain a multi-dimensional image combination.

[0034] like Figure 2 As shown, the specific detection method for edge detection is as follows:

[0035] The first step is to perform Gaussian filtering on the horizontal and vertical images.

[0036] The second step involves using edge detection operators to highlight the local edges in the horizontal and vertical images obtained in the first step. No edge detection algorithm can perfectly process the original image data; therefore, noise filtering is necessary when using edge detection operators to extract edges. This embodiment uses Gaussian filtering to smooth the image.

[0037] Furthermore, the edge detection operator is the Canny operator, and its specific process is as follows: Figure 3 As shown, when using the Canny operator to extract image edges, the gradient direction and magnitude of each pixel in the image need to be calculated after smoothing. After obtaining the gradient data, it is found that the edges extracted based on the gradient data are still not particularly clear because parasitic responses occur throughout the process. To reduce the impact of this interference on the edge detection results, non-maximum suppression is needed to filter some of the detected edge pixels.

[0038] The third step is to define the edge intensity of the pixels and extract the edge point sets of the horizontal and vertical images obtained in the second step by thresholding.

[0039] The fourth step is to determine the final basic unit edge position based on the edge point set and the distribution of adjacent edge pixels, and then draw the edge path image based on the basic unit edge position.

[0040] Step 5: Extract the outermost edge path from the edge path image as the target edge, and remove other edge paths within the target edge to obtain the horizontal and vertical target images.

[0041] By performing edge detection processing on the image, the edge position of the rolled piece in the original image (i.e., the horizontal and vertical images) can be obtained, and the black edges in the rolled piece image can be hidden, which can effectively avoid the problem of black edges interfering with subsequent similarity comparison. This solves the problem that the detection signal is easily interrupted when detecting rolled pieces at the water tank outlet through sensing components in the existing technology.

[0042] In step two, the specific method of image combination is to merge the horizontal target image into the vertical target image, ensuring that the target edges in the two images are independent of each other. This combines the two target images into a single image whose content does not interfere with each other, which is beneficial for subsequent similarity comparison operations and can greatly improve the efficiency of similarity comparison.

[0043] Step 3: Select the image with the largest horizontal area from all multi-dimensional image combinations as the baseline image combination. The image area is the area enclosed by all target edges in the multi-dimensional image combination. A larger image area indicates a more complete image, therefore using it as the baseline image combination better represents the basic formation of the rolled piece image. Furthermore, the baseline image combination obtained in this way requires no additional system settings, is applicable to rolled pieces of different types / shapes, and can effectively improve work efficiency.

[0044] Step 4: Perform similarity analysis on the baseline image combination with other multi-dimensional image combinations to extract approximate image combinations from the multi-dimensional image combinations. Specifically, the similarity analysis involves setting a similarity threshold. If the contour similarity between the baseline image combination and the multi-dimensional image combination is greater than the similarity threshold, then the multi-dimensional image combination is similar to the baseline image combination; otherwise, the multi-dimensional image combination is considered an approximate image combination. Considering the high synchronization of the rolled pieces between the water tanks in the double-height line, in this embodiment, the similarity threshold can be set to a minimum of 95%, but should not exceed 98%, thus improving the synchronization of tracking the rolled pieces in each water tank.

[0045] Step 5: If the number of approximate image combinations is less than or equal to the set counting threshold (e.g., the counting threshold is set to 1), activate the early warning system of the tank corresponding to the approximate image combination and allow all rolled pieces to be synchronously output to the tank outlet. Activating the early warning system alerts staff to the risk of asynchronous operation of the rolled pieces in the corresponding tank. If the number of approximate image combinations exceeds the set counting threshold (i.e., there are two or more approximate image combinations), the image with the second largest horizontal area among the multi-dimensional image combinations is taken as the baseline image combination, and the process returns to Step 4 to re-perform the similarity analysis and count threshold comparison. This effectively improves the accuracy of image comparison. However, if, after using the second largest multi-dimensional image combination as the baseline image combination in Steps 4 and 5, the number of approximate image combinations still exceeds the set counting threshold (i.e., there are still more than or equal to 2 approximate image combinations), it indicates a serious asynchronous operation of the rolled pieces at the tank outlet. In this case, a fault alarm is triggered, and the conveying of rolled pieces at the tank outlet is stopped.

[0046] A multi-dimensional identification system for rolled pieces at the outlet of a water tank includes a horizontal image acquisition device, a vertical image acquisition device, and a controller. The controller is electrically connected to both the horizontal and vertical image acquisition devices. The controller processes the images acquired by the horizontal and vertical image acquisition devices using the aforementioned multi-dimensional identification method to avoid interference from black edges on the rolled pieces on the detection signal, ensuring the synchronization of rolled pieces at the outlets of each water tank. This solves the problem in the prior art where detecting rolled pieces at the outlet of a water tank using sensing components easily leads to signal interruption.

[0047] The above description is only a preferred embodiment of the present invention. It should be noted that those skilled in the art can make several modifications and improvements without departing from the structure of the present invention, and these will not affect the effectiveness of the implementation of the present invention or the practicality of the patent.

Claims

1. A multi-dimensional identification method for rolled pieces at the outlet of a water tank, characterized in that, Includes the following steps: Step 1: When any sensing component is triggered, acquire horizontal and vertical images of the rolled pieces at all water tank outlets along the double-height line. Step 2: Perform edge detection and image combination processing on the horizontal and vertical images to obtain a multi-dimensional image combination; Step 3: Select the image with the largest horizontal area from all the multi-dimensional image combinations as the reference image combination; Step 4: Perform similarity analysis on the baseline image combination with other multi-dimensional image combinations to extract approximate image combinations from the multi-dimensional image combinations; Specifically, the similarity analysis includes: A similarity threshold is set. If the contour similarity between the baseline image combination and the multi-dimensional image combination is greater than the similarity threshold, then the multi-dimensional image combination is similar to the baseline image combination; otherwise, the multi-dimensional image combination is regarded as an approximate image combination. Step 5: If the number of approximate image combinations is less than or equal to a set counting threshold, activate the early warning system of the water tank corresponding to the approximate image combination; If the number of approximate image combinations exceeds the set counting threshold, then the image with the second largest horizontal area in the multi-dimensional image combination is taken as the benchmark image combination, and the process returns to step four. If, after using the second-largest multi-dimensional image combination as the reference image combination for steps four and five, the number of approximate image combinations exceeds the set counting threshold, a fault alarm will be triggered, and the conveying of the rolled piece at the water tank outlet will be stopped.

2. The multi-dimensional identification method for rolled pieces at the outlet of a water tank according to claim 1, characterized in that, The specific detection method for the edge detection is as follows: The first step is to perform Gaussian filtering on the horizontal and vertical images; The second step is to use edge detection operators to highlight the local edges of the horizontal and vertical images obtained in the first step. The third step is to define the edge intensity of the pixels and extract the edge point sets of the horizontal and vertical images obtained in the second step by threshold segmentation. Step 4: Based on the edge point set, determine the final basic unit edge position according to the distribution of adjacent edge pixels, and draw the edge path image according to the basic unit edge position; Step 5: Extract the outermost edge path from the edge path image as the target edge, and remove other edge paths within the target edge to obtain the horizontal target image and the vertical target image.

3. The multi-dimensional identification method for rolled pieces at the outlet of a water tank according to claim 2, characterized in that, The edge detection operator is the Canny operator.

4. The multi-dimensional identification method for rolled pieces at the outlet of a water tank according to claim 2, characterized in that, The specific combination method of the image combination is as follows: The horizontal target image is incorporated into the vertical target image, and the target edges in the two target images are made independent of each other.

5. The multi-dimensional identification method for rolled pieces at the outlet of a water tank according to claim 4, characterized in that, The image area is the area enclosed by all target edges in the multi-dimensional image combination.

6. A multi-dimensional identification system for rolled pieces at the outlet of a water tank, characterized in that, It includes a horizontal image acquisition device, a vertical image acquisition device, and a controller; the controller is electrically connected to both the horizontal and vertical image acquisition devices, and the controller processes the images acquired by the horizontal and vertical image acquisition devices using the multi-dimensional recognition method as described in any one of claims 1-5.

Images