A method for ultrasonic image defect detection based on image entropy and recursive analysis

Abstract

The present invention discloses an ultrasonic image defect detection method based on image entropy and recursive analysis, and belongs to the technical field of ultrasonic nondestructive testing. The method comprises: 1. performing linear array phased array ultrasonic scanning on the same workpiece or the same batch of the same workpieces, and continuously acquiring multiple B-scan images; 2. graying and median filtering preprocessing the acquired B-scan images; 3. calculating the image entropy of the preprocessed B-scan images to obtain a B-scan image entropy sequence; 4. calculating and selecting the time delay and embedding dimension of the image entropy sequence, and using them as parameters for recursive analysis; 5. recursively analyzing the image entropy sequence according to the selected parameters to obtain a recursive graph of the sequence, and judging whether there is a B-scan image containing defects in the acquired image according to the recursive graph. The present invention can effectively identify defects in multiple B-scan images, does not require image interception and segmentation operations, has a small amount of stored data, high recognition efficiency, and strong applicability.

Description

Technical Field

[0001] The present invention is applied to the technical field of ultrasonic nondestructive testing, and in particular relates to an ultrasonic B-scan image defect detection method based on image entropy and recursive analysis technology. Background Art

[0002] Ultrasonic testing is a commonly used non-destructive testing technology. Through the reflection, transmission and other properties of ultrasound, the time and amplitude of the echo are used to detect and locate defects. Ultrasonic non-destructive testing technology has become one of the important means of defect detection for extruded products because of its safe operation, accurate defect location, high sensitivity, low cost and fast speed. The phased array ultrasonic probe is composed of multiple independent chips. By controlling the excitation sequence of the chips to adjust the direction of the sound beam and the position of the focus, the B-scan image of the workpiece can be obtained. Compared with the single-chip probe, the phased array probe has high scanning efficiency and intuitive defect display, so it has been widely used in industry. In industrial inspection, workers often use handheld phased array probes to inspect workpieces. However, manual inspection has the disadvantages of low inspection efficiency, easy fatigue and missed inspection during large-scale inspection, and the judgment of defects depends on the experience of the inspection personnel. Therefore, many factories have begun to adopt phased array automatic inspection technology, relying on industrial computers to process and analyze B-scan images to identify whether there are defects.

[0003] However, when performing defect identification, false detection may occur due to noise in a single image. Multiple images should be combined for defect identification, which requires saving a large number of collected B-scan images, which also requires a large amount of storage space. When identifying defects, it is still necessary to first process and analyze the images one by one, and then combine the analysis results of multiple images to determine whether there are defects. This method has a large amount of calculation, low efficiency, and takes up a lot of storage space, so a more concise and efficient defect identification method is needed.

[0004] Image entropy can be used to evaluate the complexity of an image. For a B-scan image, if there is no defect in the workpiece, the ultrasonic image will only show the surface echo and the bottom wave, and the image complexity is relatively small. If there is a defect in the workpiece, a defect echo will appear in the area between the surface echo and the bottom wave, and the bottom wave will also be broken, causing the image complexity to increase. Image entropy can be used as the characteristic value of the B-scan image to determine whether there is a defect. Summary of the invention

[0005] The purpose of the present invention is to overcome the shortcomings of low efficiency and large data volume of traditional ultrasonic B-scan image defect recognition methods, and to provide an ultrasonic B-scan image defect detection method based on image entropy and recursive analysis technology, which can realize the recognition of defects in a large number of B-scan images.

[0006] The technical solution of the present invention is as follows:

[0007] An ultrasonic B-scan image defect detection method based on image entropy and recursive analysis technology, characterized in that the method comprises:

[0008] Step 1, performing linear array phased array ultrasonic scanning on the same workpiece or the same batch of workpieces, and continuously acquiring multiple B-scan images;

[0009] Step 2: grayscale and median filter preprocessing operations are performed on the collected B-scan image;

[0010] Step 3, calculating the image entropy of each B-scan image after preprocessing to obtain a B-scan image entropy sequence;

[0011] Step 4: Calculate and select the time delay and embedding dimension of the image entropy sequence as parameters used in the recursive analysis;

[0012] Step 5: recursively analyze the image entropy sequence according to the selected parameters to obtain a recursive graph of the image entropy sequence, determine whether there is a B-scan image containing defects in the acquired image sequence according to the recursive graph, and determine the location of the defect in the workpiece through the image with defects.

[0013] Specifically, in step 1, the workpiece is placed flat and rotated clockwise, the probe is fixed above the workpiece, and the probe is turned on. The probe continuously collects B-scan images during the rotation of the workpiece.

[0014] Specifically, in step 4, the time delay is selected by the average mutual information method, and the embedding dimension is selected by the false neighbor method.

[0015] Specifically, in step 5, if white cross lines appear in the recursive graph, it means that there are defective images among the multiple acquired B-scan images, otherwise it means that there are no defective images.

[0016] The present invention can not only use images collected from a single workpiece for processing, but also can simultaneously process images of parts of the same batch with the same size and material in practical applications. It can judge the scanned images with defects according to the coordinates of the white cross lines in the recursive graph, and determine the workpiece and position to which the defects belong according to the sequence number of the image.

[0017] By adopting the above technical solution, the beneficial effects of the present invention are:

[0018] (1) The present invention uses image entropy as the characteristic value of the B-scan image, which can not only identify the core defect echo, but also the change of the bottom wave, without the need to segment the image, and is not affected by image translation and rotation;

[0019] (2) The present invention only needs to save the image entropy of the B-scan image instead of all images, which can reduce the amount of data;

[0020] (3) The present invention recursively analyzes the image entropy sequence to detect abnormal values ​​in the sequence, rather than simply setting a threshold, which can prevent the influence of changes in workpieces or phased array parameters on image entropy;

[0021] (4) The present invention adopts recursive analysis technology to analyze the image entropy sequence, which can effectively detect B-scan images containing defects. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 The figure is a flow chart of the ultrasonic B-scan image defect detection method based on image entropy and recursive analysis technology of the present invention.

[0023] Figure 2 It is a schematic diagram of the probe scanning the rotating workpiece in the embodiment.

[0024] Figure 3 It is the B-scan image preprocessed in the embodiment.

[0025] Figure 4 is the image entropy sequence of 3200 B-scan images in the embodiment.

[0026] Figure 5 It is a recursive graph of the image entropy sequence in the embodiment.

[0027] Figure 6 These are the 2850th and 3050th B-scan images in the embodiment. DETAILED DESCRIPTION

[0028] In order to make the advantages of the invention more clear and understandable, the present invention is further described in detail below with reference to the accompanying drawings, taking the ultrasonic phased array detection of aluminum rings as an example. The specific implementation methods of the present invention are as follows:

[0029] A method for ultrasonic B-scan image defect detection based on image entropy and recursive analysis technology, such as Figure 1 As shown, the method comprises the following steps:

[0030] Step 1: Scan the aluminum ring using a linear array ultrasonic phased array, and continuously collect 3200 B-scan images. The specific scanning method is: the aluminum ring is placed flat and rotated clockwise, the probe is fixed above the aluminum ring, and the probe is turned on. The probe continuously collects B-scan images during the rotation of the aluminum ring. The diameter of the aluminum ring is 300 mm, and an image is scanned approximately every 0.3 mm. Figure 2 As shown (in order to intuitively represent the continuous acquisition of multiple images, multiple probes are drawn in the figure, but in actual operation, the workpiece rotates clockwise and the probe is stationary);

[0031] Step 2: grayscale and median filter preprocessing operations are performed on the collected B-scan image. The processing results are as follows: Figure 3As shown;

[0032] Step 3: Calculate the image entropy of the preprocessed B-scan image to obtain the B-scan image entropy sequence, such as Figure 4 As shown, we can see that the entropy values ​​at 2850 and 3050 are slightly higher than those at other parts, but the difference is very small, making it difficult to determine whether there is a defect.

[0033] Step 4, import the image entropy sequence into the recursive analysis program of MATLAB, use the average mutual information method and the false neighbor method, select the embedding dimension as 7 and the time delay as 5 as the recursive analysis parameters;

[0034] Step 5: According to the selected parameters, the recursive graph of the entropy sequence is obtained as follows: Figure 5 As shown in the figure, it can be seen that there are white cross lines at coordinates 2850 and 3050 in the recursive graph, indicating that there are defects in the 2850th and 3050th B-scan images. The corresponding position on the workpiece can be determined by the number of the defective B-scan image, such as Figure 2 shown.

[0035] The two images with serial numbers 2850 and 3050 are as follows Figure 6 As shown, there is an obvious defect echo. After destructive verification, it was found that the aluminum ring has a defect at the position corresponding to the image.

[0036] The present invention uses image entropy to replace the entire image, which can effectively reduce the amount of data; recursive analysis is performed on the image entropy sequence of multiple images to detect abnormal values ​​therein, rather than simply setting a threshold, which can prevent the influence of changes in workpieces or phased array parameters on image entropy.

[0037] This embodiment uses images collected by a single aluminum ring for processing. In practical applications, images of parts of the same batch and the same size and material can be processed simultaneously. The coordinates of the white cross lines in the recursive graph can be used to determine the defective workpiece and the location of the defect on the workpiece.

[0038] The above-described embodiments are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the scope of protection determined by the claims of the present invention.

Claims

1. A method for ultrasonic image defect detection based on image entropy and recursive analysis, characterized in that: The method includes: Step 1, performing linear array phased array ultrasonic scanning on the same workpiece or a batch of workpieces with the same specifications, and continuously acquiring multiple B-scan images; Step 2: grayscale and median filter preprocessing operations are performed on the collected B-scan image; Step 3, calculating the image entropy of each B-scan image after preprocessing to obtain a B-scan image entropy sequence; Step 4: Calculate and select the time delay and embedding dimension of the image entropy sequence as parameters used in the recursive analysis; Step 5: recursively analyze the image entropy sequence according to the selected parameters to obtain a recursive graph of the image entropy sequence, determine whether there is a B-scan image containing defects in the collected image sequence according to the recursive graph, and determine the location of the defect in the workpiece through the image with defects.

2. The ultrasonic image defect detection method based on image entropy and recursive analysis according to claim 1, characterized in that: In step 4, the time delay is selected by the average mutual information method, and the embedding dimension is selected by the false neighbor method.

3. The ultrasonic image defect detection method based on image entropy and recursive analysis according to claim 1, characterized in that: In step 5, if white cross lines appear in the recursive graph, it means that there are defective images among the multiple B-scan images collected, otherwise it means that there are no defective images.

Images