TW201737857A - Image processing method - Google Patents

Abstract

An image processing method is used to obtain a boundary range. The method includes sampling an image from an inner side of an object; capturing a first boundary curve from the image; capturing a plurality of preliminary boundary curves from the image; obtaining a first reference curve according to the preliminary boundary curves; obtaining a thickness according to the first reference curve and the first boundary curve; obtaining a plurality of candidate boundary curves by processing the preliminary boundary curves according to the first boundary curve and the thickness; obtaining a second reference curve according to the candidate boundary curves; selecting a set of nodes from the candidate boundary curves according to the second reference curve; connecting the set of nodes to form a resultant boundary curve; and defining the boundary range according to the first boundary curve and the resultant boundary curve.

Description

Image processing method

The invention discloses an image processing method, in particular to an image processing method capable of obtaining a result boundary curve according to a plurality of initial boundary curves to define a boundary range.

The use of energy waves to detect the interior of an object is a common application in engineering or medical fields. Mechanical waves (such as ultrasonic waves) or electromagnetic waves (such as X-rays) are commonly used to detect the inside of an object and to be imaged for inspection.

However, it is not easy to define a boundary curve in a blurred image with an automated program. For medical applications, for example, cardiovascular physicians often use medical ultrasound to detect the patient's neck to detect the intima-media thickness (IMT) boundary of the blood vessel. If the thickness is too thick, it can warn that the risk of cardiovascular disease is too high. However, ultrasound imaging is often blurred after imaging, so it is highly dependent on manual interpretation by relevant personnel (such as physicians, inspectors, engineers, etc.). If manual interpretation is not used, it is not easy to define the boundary range to be detected. Under the premise of manual interpretation, even if there is original image, it is not easy to use image processing method to know the boundary curve in the blurred original image. This situation not only causes the burden of manpower demand, but also makes it difficult to realize a large amount of data analysis.

Therefore, there is a need in the art for a solution to help the relevant personnel to more easily define the boundary range to be detected and to improve the feasibility and accuracy of automated analysis.

The invention provides an image processing method for determining a boundary range. The method includes: sampling an image of an inner wall of an object; extracting a first boundary curve from the image; extracting a plurality of initial boundary curves from the image; and obtaining a first reference curve according to the initial boundary curves; Determining a thickness of the first reference curve and the first boundary curve; processing the initial boundary curves according to the first boundary curve and the thickness to obtain a plurality of candidate boundary curves; obtaining a first according to the candidate boundary curves a second reference curve; selecting a set of nodes from the candidate boundary curves according to the second reference curve; connecting the set of nodes to form a result boundary curve; and defining the boundary range according to the first boundary curve and the result boundary curve Wherein the first boundary curve corresponds to a first side of the boundary range, and the initial boundary curve, the candidate boundary curve, and the result boundary curve correspond to a second side of the boundary range.

The use of ultrasonic waves to detect blood vessels is mentioned below as an example to illustrate the principles of embodiments of the present invention. However, the image processing method of the present invention is not limited to this application, and other applications and reasonable variations of the image processing method using the present invention are still within the scope of the present invention.

Fig. 1 is a view of an ultrasonic image 100 of an inner wall of a blood vessel observed in an embodiment of the present invention. The blood vessel includes an upper vessel wall 110, a lower vessel wall 120, and a lumen 130 for blood circulation. Between the two arrows shown in Fig. 1, it can be the inner media thickness (IMT) to be observed, that is, the boundary range to be defined.

2 is a schematic diagram of a sample image 210 in an embodiment of the present invention. As described above, if the method provided by the embodiment of the present invention is not used, it is necessary to rely on manual interpretation to know the boundary range to be defined (for example, the thickness of the media in the blood vessel), so that it is not only labor-intensive, but also is not conducive to improving the automatic interpretation. The correctness. When the method of the embodiment of the present invention is used, the image 210 can be sampled at the boundary between the lower blood vessel wall 120 and the inner cavity 130, and the first boundary curve 220 can be extracted from the image 210. In this example, the first boundary curve 220 can be a lumen-intima interface (LII) curve of the underlying vessel wall 120. Also seen in FIG. 2 is a second boundary curve 230, which may be a media-adventitia interface (MAI) curve of the lower vessel wall 120. The boundary range between the first boundary curve 220 and the second boundary curve 230 may be the aforementioned inner media thickness.

As can be seen from Fig. 2, since the gray scale of the inner cavity after imaging is low (black) and the gray scale of the inner membrane is high (white), the first boundary curve 220 can be determined more clearly. However, since the gray scale of the middle membrane and the outer membrane is similar, the second boundary curve 230 is easily affected by the image of the thickness of the inner membrane and is less identifiable, and can be clearly defined by further processing. .

FIG. 3 is a flow chart of an image processing method 300 according to an embodiment of the present invention. According to Figures 1 and 2, the image processing method 300 is used to determine the intima-media thickness of the blood vessel, which may include the following steps, wherein step 330 may correspond to Figures 4 to 9, and steps 340 to 360 may correspond to Figure 10. Steps 370 to 390 may correspond to Figure 11, and the steps will be detailed below:

Step 310: sampling image 210 at the junction of the lower vessel wall 120 and the lumen 130;

Step 320: Draw a first boundary curve 220 from the image 210;

Step 330: Extract a plurality of initial boundary curves 2301-230x from the image 220;

Step 340: Determine a first reference curve 240 according to a plurality of initial boundary curves 2301-230x;

Step 350: Determine the thickness TH according to the first reference curve 240 and the first boundary curve 220;

Step 360: processing a plurality of initial boundary curves 2301-230x according to the first boundary curve 220 and the thickness TH to obtain a plurality of candidate boundary curves 2301'-230x';

Step 370: Determine a second reference curve 235 according to the plurality of candidate boundary curves 2301'-230x';

Step 380: Select a group of nodes P1-Pk from a plurality of candidate boundary curves 2301'-230x' according to the second reference curve 235;

Step 390: Connect the nodes P1-Pk to form a result boundary curve 239; and

Step 395: Defining the boundary range according to the first boundary curve 220 and the result boundary curve 239 to determine the inner media thickness.

Wherein, the first boundary curve 220 may correspond to a first side of the boundary range, such as, but not limited to, an upper side. The initial boundary curves 2301-230x, the candidate boundary curves 2301'-230x', and the resulting boundary curves may correspond to a second side of the boundary range, such as, but not limited to, the underside. The result boundary curve obtained by the above steps may be the second boundary curve 230 shown in FIG. 2, and if the application of measuring the inner film thickness of the blood vessel is taken as an example, the middle membrane-outer membrane interface (MAI) may be used. )curve. The initial boundary curve and the candidate boundary curve described in the above steps 330 to 380 are numbered x. The number of nodes P1-Pk in step 380 is k. x, k can be positive integers greater than 1, and can be adjusted according to engineering needs. The principle of the embodiment of the present invention will be described below by taking x=3 as an example.

4 is a flow chart of a method for extracting an initial boundary curve 2301 in an embodiment of the present invention. The initial boundary curve 2301 described in step 330 of FIG. 3 can be obtained, for example, by the following steps:

Step 4301: Enhance the contrast of the image 210 to form a second image 2102;

Step 4302: Perform a smoothing filter process on the second image 2101 to form a third image 2103;

Step 4303: Perform binarization processing on the third image 2103 to form a fourth image;

Step 4304: Perform a morphological process on the fourth image to form a fifth image 2105; and

Step 4305: Capture an initial boundary curve 2301 according to the fifth image 2015.

Fig. 5 is a diagram showing an image processing change corresponding to each step of Fig. 4. As can be seen from the distribution of the image of the image 210, the line segment 210a is essentially the intermediate film-outer film interface (MAI) curve. Step 4301 improves the contrast of the image 210 to facilitate subsequent image processing. The smoothing filtering process of step 4302 can be, for example, a Gaussian filter or a bilateral filter process, which can reduce de-noise and make the image more uniform and smooth. The binarization process of step 4303 can convert the image from a grayscale image to a black and white image, which can facilitate processing the boundary portion of the image. The morphological processing described in step 4304 may include a dilation process and/or an erosion process, both of which are exemplified herein. The expansion process can fill and eliminate the dark spots of the high gray scale (such as the white part) by the expansion method, and the etching process can retract the layout of the image back to the pattern before the expansion process, so that the effect of filling the void can be filled. The second image 2103 is subjected to binarization processing and expansion processing to generate an image 2105'. The image 2105' is etched to produce a fifth image 2105, and an initial boundary curve 2301 is taken accordingly. The initial boundary curve 2301 can be superimposed back onto the image 210 for comparison.

The portion above the line segment 2105a of the image 2105' and the fifth image 2105 may substantially correspond to the upper portion of the film-outer film interface (MAI) curve, so after processing, it should be displayed as a black region, but the second view. In the image 2102 and the third image 2103, the film-outer film interface (MAI) curve in the left half is less clear, so after binarization (step 4303) and morphological processing (step 4304), above the line segment 2105a Some parts are shown as white areas. Therefore, the initial boundary curve 2301 taken from the boundary between the black area and the white area of the fifth image 2015 in step 3305 is as shown in FIG. 5, and the line segment 210a (which may substantially correspond to the middle film sought) The outer membrane interface curve is inconsistent and has errors. In this example, especially in the left half of the image, this error is significant. Therefore, in accordance with an embodiment of the present invention, in addition to the initial boundary curve 2301, other initial boundary curves, such as 2302, 2303, must be derived to obtain a more accurate boundary range (the boundary range of this example, ie, the intima thickness).

Figure 6 is a flow chart of a method for extracting an initial boundary curve 2302 in an embodiment of the present invention. The initial boundary curve 2302 described in step 330 of FIG. 3 can be obtained, for example, by the following steps:

Step 4301: Enhance the contrast of the image 210 to form a second image 2102;

Step 6302: Perform a filtering process on the second image 2102 to form a sixth image 7106;

Step 6303: Perform image enhancement on the sixth image to form a seventh image 7107; and

Step 6304: The initial boundary curve 2302 is captured according to the seventh image 7107.

Fig. 7 is a diagram showing an image processing change corresponding to each step of Fig. 6. The filtering process described in step 6302 may include, for example, a medium filter process and/or an edge filter process, where both are taken as an example. Median filtering can be used to eliminate speckle noise. The edge filtering process can detect the edge position in the image. Here, for example, Sobel filter processing can be used as the calculation method of the edge filtering process. In Fig. 7, a median filtering process is performed on the second image 2102 to generate an image 7106', and an edge filtering process is performed on the image 7106' to generate a sixth image 7106. In step 6303, image enhancement is performed on the sixth image 7106 to make the boundary of the image more clear, and a seventh image 7107 is generated. In step 6304, an initial boundary curve 2302 can be captured in the seventh image 7107. An initial boundary curve 2302 can be used to overlay the image 210. As shown in Fig. 5, in the left half of the image, the initial boundary curve 2301 may be inaccurate with the line segment 2105a which is closer to the film-outer film interface curve, so the steps of steps 6 and 7 are used. The initial boundary curve 2302, and the initial boundary curve 2303 described below, can compensate for the error between the line segment 2105a of the fifth image 2105 and the initial boundary curve 2301.

Figure 8 is a flow chart of a method for extracting an initial boundary curve 2303 in an embodiment of the present invention. The initial boundary curve 2303 described in step 330 of FIG. 3 can be obtained, for example, by the following steps:

Step 4301: Enhance the contrast of the image 210 to form a second image 2102;

Step 6302: Perform a filtering process on the second image 2102 to form a sixth image 7106;

Step 6303: performing image enhancement on the sixth image to form a seventh image 7107;

Step 8304: Perform a binarization process on the seventh image 7107 to form an eighth image 8108;

Step 8305: Perform morphological processing on the eighth image 8108 to form a ninth image 8109; and

Step 8306: The initial boundary curve 2303 is captured according to the ninth image 8109.

Fig. 9 is a diagram showing an image processing change corresponding to each step of Fig. 8. The description of steps 4301, 6302 and 6303 is as above and will not be described again. In step 8304, performing binarization processing on the seventh image 7107 can highlight edge portions in the image. The morphological treatment of step 8305 may comprise an etch treatment and/or an expansion treatment, both of which are exemplified herein. In step 8305, after the etching process is performed on the eighth image 8108, the image 8109' can be generated. This etching process can eliminate the noise portion of the low gray level region (such as the black region), thereby facilitating the correctness of the subsequent image processing. Performing the expansion process on the image 8109' can restore the reduced area of the etching process, thereby producing a ninth image 8109. In step 8306, the upper edge of the white region of the ninth image 8109 can approximately correspond to the desired film-outer film interface (MAI) curve, so the initial boundary curve 2303 can be extracted from the ninth image 8109.

As shown in Figures 4 through 9, initial boundary curves 2301, 2302, 2303 can be found. In step 340 of FIG. 3, the first reference curve 240 can be found from the initial boundary curves 2301 to 230x (here, 2301 to 2303 are taken as an example). According to an embodiment of the present invention, the average calculation may be performed according to the position information of the initial boundary curves 2301, 2302, and 2303, and the obtained average curve may be the first reference curve 240. This average calculation can use arithmetic averaging, weighted averaging, geometric averaging, or other calculation functions. The position information may be, for example, a coordinate value of the longitudinal axis direction of the initial boundary curves 2301 to 2303. The longitudinal axis direction may be substantially perpendicular to the first side (upper side) and the second side (such as the side) of the image 210. FIG. 10 is a schematic diagram showing the initial boundary curves 2301, 2302, and 2303 placed in the sampled image 210 in the embodiment of the present invention. After averaging, a first reference curve 240 can be obtained, and the first reference curve 240 and the first boundary curve 220 can be a thickness TH.

In step 360, the initial boundary curve 2301-230x is processed according to the first boundary curve 220 and the thickness TH to obtain a candidate boundary curve, for example, an allowable range is defined according to the first boundary curve 220 and the thickness TH, and the initial boundary curve 2301 is adjusted. The portion of the -230x that exceeds the allowable range is obtained to find the candidate boundary curves 2301'-230x'. For example, a band-shaped region formed by the thickness TH below the first boundary curve 220 may be defined as an allowable range, if n pixels (n is a positive integer) on the initial boundary curve 2301 fall outside the allowable range. Then, the n pixels of the initial boundary curve 2301 can be deleted, or the curve can be adjusted to pull the portion of the curve falling outside the allowable range into the allowable range, thereby generating the candidate boundary curve 2301'. According to an embodiment of the invention, for example, the thickness TH below the first boundary curve 220 may be multiplied by a parameter or corrected by other mathematical functions to generate the allowable range.

According to an embodiment of the present invention, in step 360, the allowable range may be defined according to the first boundary curve 220, the thickness TH, and a threshold value, so as to adjust a portion of the initial boundary curve 2301-230x that exceeds the allowable range, thereby Candidate boundary curves 2301'-230x' are obtained. The threshold value may be, for example, a band boundary formed by the first boundary curve 220 and the thickness TH, a lower m pixels (m-type positive integer), or a band above the next distance, and the distance is The k% of the thickness TH (100 ≤ k < 0) and the like.

Step 360 may remove the portion of the initial boundary curve 2301-230x that is too offset to cause the candidate boundary curve 2301'-230x' to fall within a reasonably tolerable range. In step 370, an average calculation (which may be an arithmetic average, a weighted average, a geometric mean, or other calculation function) may be performed using candidate boundary curves 2301'-230x', and the resulting average curve may be a second reference curve 235.

Figure 11 is a schematic illustration of the principle of step 380 of Figure 3. Step 380 refers to candidate boundary curves 2301' to 230x', and in Fig. 11, the candidate boundary curves 2301' to 2303' are taken as an example. Figure 11 is a schematic diagram showing the operation of selecting a set of nodes P1-P18 according to the second reference curve 235 and the candidate boundary curves 2301'-2303' in the embodiment of the present invention. The second reference curve 235 has approximated the sought second boundary curve 230 (eg, the mid-membrane-outer membrane interface MAI curve for vascular ultrasound detection), but since the second reference curve 235 is from the candidate boundary curve 2301'-2303 'The average calculation is performed, so it is different from the curve directly measured by the image. Therefore, in order to bring the final result closer to the curve directly measured by the image, the operation of Fig. 11 can be performed. Step 380 can include:

Step 3801: The plurality of axes A1-Ak are drawn on the candidate boundary curves 2301'-230x', and the axes A1-Ak are substantially perpendicular to the first side and the second side, thereby making the axes A1 -Ak and the candidate boundary curves 2301'-230x' form a set of candidate nodes; and

Step 3802: Pick the candidate nodes on each of the axes A1-Ak that are closest to the second reference curve 235 to form the set of nodes P1-Pk.

For example, x = 3 and k = 18 are illustrated, but the embodiment of the present invention is not limited to this aspect. The candidate nodes may be a plurality of boundary points of the axis A1-A18 and the candidate boundary curves 2301'-230x'. In Fig. 11, there may be 3 candidate nodes on each axis. With regard to step 3802, taking the axis A1 as an example, among the candidate nodes on the axis A1, the boundary point between the candidate boundary curve 2301' and the axis A1 is closest to the second reference curve 235, so the boundary point between the candidate boundary curve 2301' and the axis A1 It can be selected as the node P1 on the axis A1; further, the boundary point of the candidate boundary curve 2302' and the axis A2 is closest to the second reference curve 235 on the axis A2, so the boundary point between the candidate boundary curve 2302' and the axis A2 can be Selected as node P2 on axis A2, and so on, nodes P1-P18 can be selected on axes A1-A18. As described in step 390, nodes P1-P18 are connected to form a resulting boundary curve 239.

The result boundary curve 239 obtained by the principle of Fig. 11 can correspond to the second boundary curve 230 of Fig. 2. If the application of the vascular ultrasound test is taken as an example, the result boundary curve 239 can be difficult for the prior art. The mid-membrane-outer membrane interface (MAI) curve is discriminated, but can be defined using the method of the embodiments of the present invention. In Fig. 11, since the number of axes A1-A18 is only 18, the resulting boundary curve 239 is micro-serrated, however, Fig. 11 is only an example for explaining the principle of operation, when the number of axes is sufficient, That is, when the sampling rate is high and the resolution of the display device is sufficient, the resulting boundary curve 239 can be a more detailed curve.

FIG. 12 is a flow chart showing a method of extracting the first boundary curve 220 in step 320 of FIG. 3 in an embodiment of the present invention. Fig. 13 is a diagram showing an image processing change corresponding to Fig. 12. Step 320 can include:

Step 3201: The contrast of the image 210 is lowered to form an image 1301; steps 3202 and 3206 are entered;

Step 3202: Perform smoothing filtering on the image 1301 to form an image 1302;

Step 3203: Perform image edge enhancement on image 1302 to form image 1303;

Step 3204: Perform image binarization and expansion processing on image 1303 to form image 1304;

Step 3205: Perform an etching process on the image 1304 to form an image 1305; proceed to step 3208;

Step 3206: Perform image filtering processing on image 1301 to form image 1306;

Step 3207: Perform image detection on the image 1306 to obtain the threshold line 1380; proceed to step 3209;

Step 3208: Capture the first boundary curve 220 according to the image 1305; go to step 3209;

Step 3209: Is the first boundary curve 220 within the range of the threshold line 1380? If yes, go to step 3210; if no, go to step 3211;

Step 3210: The first boundary curve 220 is within a reasonable range, showing a first boundary curve 220;

Step 3211: The first boundary curve 220 is outside the reasonable range, and the error message is reported.

The smoothing filtering process of step 3201 may be, for example, Gaussian filtering or bidirectional filtering processing. The edge filtering processing described in step 3206 may be, for example, Sobel filtering processing, and various filtering processing methods and corresponding functions described in steps 3201 to 3205, 3206 may be referred to. The foregoing, so it is not repeated. Steps 3201 to 3205, 3208 may determine a first boundary curve 220, and steps 3208 through 3211 may be selectively performed to check whether the first boundary curve 220 that has been captured is within a reasonable range. In this example, the range of the threshold line 1380 described in step 3209 can be, for example, below the threshold line 1380. Taking the application of vascular ultrasound detection as an example, the first boundary curve 220 may be a lumen-intimal interface (LII) curve of the vessel wall.

Figure 14 is a schematic illustration of vascular ultrasound images 210a through 210d and corresponding first boundary curves 220a through 220d and resulting boundary curves 239a through 239d in an embodiment of the present invention. Taking the image 210a as an example, the first boundary curve 220a and the result boundary curve 239a can be obtained by using the method of the embodiment of the present invention. The first boundary curve 220a can be a lumen-intimal interface (LII) curve, and the result boundary curve 239a can be a medial-outer membrane interface (MAI) curve, and the boundary between the two curves can be the desired intima-media thickness (IMT). As can be seen from FIG. 14, the method disclosed in the embodiment of the present invention can effectively extract the interface curve from the initial image (such as the image 210a to 210d), and in particular overcome the misjudgment caused by the difficulty in defining the interface curve in the image blur. Here, the application of the vascular ultrasound is taken as an example to illustrate the principle of the present invention, but the application of the present invention is not limited to the medical field. For example, in the field of fluid analysis, meteorological or marine research, civil engineering, mechanical analysis, or other areas where image signal analysis is to be performed, the method disclosed in the embodiments of the present invention can be used to assist relevant personnel in defining interface curves in a blurred image. . In summary, the present invention is useful for processing various applications of blurred images. The above are only the preferred embodiments of the present invention, and all changes and modifications made to the scope of the present invention should be within the scope of the present invention.

100‧‧‧ image map
110‧‧‧ upper vessel wall
120‧‧‧ lower vessel wall
130‧‧‧ lumen
300‧‧‧ method
310 to 395, 4301, 6302 to 6304, 8304 to 8306, 3201 to 3211 ‧ ‧ steps
220, 220a to 220c‧‧‧ first boundary curve
230‧‧‧second boundary curve
210, 2105', 7106', 1301 to 1306, 210a to 210c‧ ‧ images
1380‧‧‧ threshold line
210a, 2105a‧‧‧ segments
2102‧‧‧Second image
2103‧‧‧ Third image
2105‧‧‧ fifth image
7106‧‧‧6th image
7107‧‧‧ seventh image
8108‧‧‧8th image
7109‧‧‧9th image
2301, 2302, 2303‧‧‧ initial boundary curve
240‧‧‧First reference curve
TH‧‧‧ thickness
A1 to A18‧‧ Axis
P1 to P18‧‧‧ nodes
2301', 2302', 2303'‧‧‧ candidate boundary curve
235‧‧‧second reference curve
239, 239a to 239c‧‧‧ Results Boundary Curve

Fig. 1 is a view showing an ultrasonic image of the inner wall of a blood vessel observed in the embodiment of the present invention. Figure 2 is a schematic diagram of a sampled image in an embodiment of the present invention. FIG. 3 is a flow chart of an image processing method according to an embodiment of the present invention. Figure 4 is a flow chart of a method for extracting an initial boundary curve in an embodiment of the present invention. Fig. 5 is a diagram showing an image processing change corresponding to each step of Fig. 4. Figure 6 is a flow chart of a method for extracting an initial boundary curve in an embodiment of the present invention. Fig. 7 is a diagram showing an image processing change corresponding to each step of Fig. 6. Figure 8 is a flow chart of a method for extracting an initial boundary curve in an embodiment of the present invention. Fig. 9 is a diagram showing an image processing change corresponding to each step of Fig. 8. Figure 10 is a schematic diagram showing the initial boundary curve placed in the sampled image in the embodiment of the present invention. Figure 11 is a schematic illustration of the principle of step 380 of Figure 3. Figure 12 is a flow chart of a method for extracting a first boundary curve in step 320 of Figure 3 in an embodiment of the present invention. Fig. 13 is a diagram showing an image processing change corresponding to Fig. 12. Figure 14 is a schematic diagram of a vascular ultrasound image and a corresponding first boundary curve and a resulting boundary curve in an embodiment of the present invention.

310 to 395 ‧ steps

Claims (12)

An image processing method for obtaining a boundary range, the method comprising: sampling an image of an inner wall of an object; extracting a first boundary curve from the image; and extracting a plurality of initial boundary curves from the image; Obtaining a first reference curve according to the initial boundary curve; determining a thickness according to the first reference curve and the first boundary curve; processing the initial boundary curves according to the first boundary curve and the thickness to obtain a plurality of a candidate boundary curve; obtaining a second reference curve according to the candidate boundary curves; selecting a set of nodes from the candidate boundary curves according to the second reference curve; connecting the set of nodes to form a result boundary curve; The first boundary curve and the result boundary curve define the boundary range; wherein the first boundary curve corresponds to a first side of the boundary range, and the initial boundary curve, the candidate boundary curve, and the result boundary The curve corresponds to a second side of the boundary range. The method of claim 1, the processing the initial boundary curves according to the first boundary curve and the thickness to obtain the candidate boundary curves, comprising: defining an allowable range according to the first boundary curve and the thickness; And adjusting a portion of the initial boundary curves that exceeds the allowable range to obtain the candidate boundary curves. The method of claim 1, the processing the initial boundary curves according to the first boundary curve and the thickness to obtain the candidate boundary curves, the method comprising: defining, according to the first boundary curve, the thickness, and a threshold value The allowable range; and adjusting portions of the initial boundary curves that exceed the allowable range to obtain the candidate boundary curves. The method of claim 1, selecting the set of nodes from the candidate boundary curves according to the second reference curve, comprising: dividing the plurality of axes onto the candidate boundary curves, wherein the axes are substantially Vertically perpendicular to the first side and the second side, such that the axes and the candidate boundary curves form a set of candidate nodes; and selecting candidate nodes on each axis that are closest to the second reference curve to form the group node. The method of claim 1, wherein the initial boundary curves comprise a first initial boundary curve; and extracting the initial boundary curves from the image comprises: increasing the contrast of the image to form a second image; Performing a smoothing process on the second image to form a third image; performing a first binarization process on the third image to form a fourth image; performing a first morphological process on the fourth image to Forming a fifth image; and extracting the first initial boundary curve according to the fifth image. The method of claim 5, wherein the first morphological treatment comprises an expansion treatment and/or an erosion treatment. The method of claim 5, wherein the initial boundary curves further comprise a second initial boundary curve; extracting the initial boundary curves from the image, comprising: performing a filtering process on the second image to form a a sixth image; performing an image enhancement on the sixth image to form a seventh image; and extracting the second initial boundary curve according to the seventh image. The method of claim 7, wherein the filtering process comprises a median filtering process and/or an edge filtering process. The method of claim 7, wherein the initial boundary curves further comprise a third initial boundary curve; and extracting the initial boundary curves from the image comprises: performing a second binarization on the seventh image Processing to form an eighth image; performing a second morphological process on the eighth image to form a ninth image; and extracting the third initial boundary curve according to the ninth image. The method of claim 9, wherein the second morphological treatment comprises an expansion treatment and/or an erosion treatment. The method of claim 1, wherein the first reference curve is an average curve obtained by performing an averaging calculation based on position information of the initial boundary curves. The method of claim 1, wherein the second reference curve is an average curve obtained by performing an averaging calculation based on position information of the candidate boundary curves.