Character recognition methods, electronic devices, and storage media in the treatment control system display

Abstract

This invention provides a character recognition method, electronic device, and storage medium for a treatment control system display. The character recognition method includes: acquiring an image from the treatment control system display; acquiring an image to be recognized containing characters; binarizing the image to be recognized; removing edge blank areas from the binarized image; performing character segmentation on the binarized image to obtain several first character image blocks, where each first character image block is either a single-character image block or a connected character image block; confirming the category of the first character image blocks and performing the following processing: if the first character image block is a single-character image block, then by matching it with a template character image, the single character in the single-character image block is recognized; if the first character image block is a connected character image block, then the connected character image block is de-connected to recognize each character in the connected character image block. This invention enables automatic recognition of planning information in a treatment control system display.

Description

Technical Field

[0001] This invention relates to the medical field, and more particularly to a character recognition method, electronic device, and storage medium in a treatment control system display. Background Technology

[0002] Before radiotherapy, it is necessary to read the patient's treatment plan information displayed on the treatment control system monitor to ensure the accuracy of the patient's information.

[0003] Currently, the treatment control system display is usually read manually to identify the plan information. However, due to eye fatigue, errors are prone to occur.

[0004] Therefore, it is necessary to use automatic image recognition technology to automatically identify planning information. However, planning information not only contains the 26 uppercase and 26 lowercase English letters, but also special characters, making recognition difficult. Furthermore, the planning information may contain connected characters, further increasing the difficulty of automatic recognition. For these reasons, using known image recognition methods to identify planning information results in a high error rate. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention first provides a character recognition method for a treatment control system display, which enables automatic recognition of planning information displayed on the treatment control system display. The detailed technical solution of the character recognition method of this invention is as follows:

[0006] A character recognition method in a treatment control system display includes:

[0007] Acquire images from the display of the treatment control system;

[0008] Obtain an image to be recognized, including characters, from a predetermined region of the image;

[0009] The image to be recognized is binarized to obtain a binary image;

[0010] Remove the blank areas at the edges of the binarized image;

[0011] Character segmentation is performed on the binarized image after removing edge blank areas to obtain several first character image blocks, where the first character image blocks are single character image blocks or connected character image blocks;

[0012] The category of the first character image block is determined, and the following processing is performed:

[0013] If the first character image block is a single character image block, then the single character in the single character image block is identified by matching it with the template character image;

[0014] If the first character image block is a contiguous character image block, then the contiguous character image block is subjected to de-contiguous processing in order to recognize each character in the contiguous character image block.

[0015] In some embodiments, character segmentation is performed on the binarized image after removing edge blank areas to obtain a plurality of first character image blocks, including:

[0016] The binarized image is segmented from left to right using connected component analysis.

[0017] Remove the blank areas at the edges of each segmented image obtained from the segmentation to obtain several first character image blocks.

[0018] In some embodiments, identifying the category of the first character image block includes:

[0019] The width of the first character image block is compared with a predetermined threshold to determine the category of the first character image block, specifically:

[0020] If the width of the first character image block is less than the predetermined threshold, the first character image block is determined to be a single character image block; otherwise, the first character image block is determined to be a connected character image block.

[0021] In some embodiments, a de-adhesion process is performed on the image blocks of adhered characters to enable the recognition of each character in the image blocks, including:

[0022] Connectivity analysis is performed on the connected character image block. Based on the high and low positions in the connected component analysis results, the connected character image block is coarsely divided into several second image character blocks. The second image character block is an uppercase single character graphic block including an uppercase character, or a lowercase single character graphic block including a lowercase single character, or an equal-height connected character image block including at least a number of uppercase single characters or a number of lowercase single characters.

[0023] Confirm the category of the second character image block and perform the following processing:

[0024] If the second character image block is an uppercase single character graphic block or a lowercase single character graphic block, then by matching it with the template character image, the uppercase single character or lowercase single character in the uppercase single character graphic block or lowercase single character graphic block is identified.

[0025] If the second character image block is a series of interconnected character image blocks of equal height, then further de-connection processing is performed on the interconnected character image blocks of equal height in order to recognize each character in the interconnected character image blocks of equal height.

[0026] In some embodiments, confirming the category of the second character image block includes:

[0027] The width of the second character image block is compared with a predetermined threshold to determine the category of the second character image block, specifically:

[0028] If the width of the second character image block is less than the predetermined threshold, the second character image block is determined as an uppercase single character graphic block or a lowercase single character graphic block; otherwise, the second character image block is determined as a character image block with equal height and adhering to each other.

[0029] In some embodiments, the process of performing de-cohesion processing on the image blocks of characters with equal height to recognize each character in the image blocks of characters with equal height includes:

[0030] The image of characters with equal height and overlapping is resized to obtain a processed image T of characters with equal height and overlapping with the target size. new ;

[0031] Based on ratio T new The template character image is resized using _h / M_h to obtain the processed character image T, which is connected to the same height. new Template character images M with the same height new , among which, T new _h represents the processed image of the contour-aligned characters T. new The height value, M_h is the height value of the template character image before processing;

[0032] The template character image M after processing the template character image new Width dimensions of the image T of the character with equal height adhesion new Perform cropping to extract image blocks containing a single third character one by one;

[0033] By matching with the template character image, the recognition of single characters in the third character image block is performed.

[0034] A second aspect of the present invention provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the character recognition method provided in the first aspect of the present invention.

[0035] A third aspect of the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the character recognition method described in the first aspect of the present invention.

[0036] The technical solution provided by this invention is based on image recognition technology, which enables automatic recognition of treatment plan information displayed on a treatment control system screen. In particular, it solves the problem of segmenting and recognizing connected characters included in the treatment plan information, thereby improving the accuracy of treatment plan information recognition. Attached Figure Description

[0037] Figure 1 A flowchart of a character recognition method provided in one embodiment of the present invention;

[0038] Figure 2 A flowchart of a character recognition method provided in another embodiment of the present invention;

[0039] Figure 3 A flowchart of a character recognition method provided in another embodiment of the present invention;

[0040] Figure 4 This is a schematic diagram illustrating the execution process of the character recognition method in this embodiment of the invention;

[0041] Figure 5 This is a schematic diagram illustrating the effects of each stage during the execution of the character recognition method in this embodiment of the invention. Detailed Implementation

[0042] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0043] In this invention, the phrase "in some embodiments" does not specifically refer to the same embodiment; they may be the same embodiment or different embodiments.

[0044] like Figure 1 As shown, the character recognition method in the display of the treatment control system provided by the present invention includes the following steps:

[0045] S100: Acquire the image displayed on the treatment control system monitor.

[0046] In practice, an image acquisition card can be used to capture images displayed on the treatment control system monitor. The images displayed on the treatment control system monitor are in a standard format, including patient treatment plan information displayed as characters. This displayed treatment plan information is located within a predetermined display area within the image, such as a predetermined-sized display area in the center of the image.

[0047] In addition, images on the treatment control system display can also be obtained by taking pictures with a camera. However, due to limitations such as shooting distance and pixel count, the characters in the obtained images are more likely to stick together.

[0048] S200: Obtain the image to be recognized, including characters, from a predetermined area of ​​the image.

[0049] The image to be identified is located within a predetermined display area of ​​the image, and the display area of ​​the image to be identified is generally preset. Therefore, it is only necessary to capture the image within this predetermined area to achieve the acquisition of the image to be identified.

[0050] S300. Perform binarization processing on the image to be recognized to obtain a binarized image.

[0051] After binarization, the image becomes simpler, the data volume is reduced, and the characters in the image to be recognized are fully highlighted. The technology for image binarization is very mature, and various known binarization algorithms can be selected as needed during the specific implementation process, which will not be elaborated on in this manual.

[0052] S400: Remove blank areas at the edges of the binarized image.

[0053] Specifically, the binarized image can be segmented horizontally and vertically to obtain a bounding rectangle containing the target character. In the binarized image obtained with this bounding rectangle as the border, the top of uppercase characters is at the same height as the top edge of the bounding rectangle, while the top of lowercase characters has a gap with the top edge of the bounding rectangle. This lays the foundation for subsequent character segmentation.

[0054] S500. Perform character segmentation on the binarized image with the edge blank areas removed to obtain several first character image blocks. The first character image blocks are single character image blocks or connected character image blocks.

[0055] Since the character pixels in a binarized image are zero, while the background pixel values ​​are non-zero, a method familiar to those skilled in the art—scanning the binary image from left to right—can be used to perform the initial character segmentation. Specifically, after the scan begins, the column where the first pixel value is zero is marked as the left boundary of the first character. Then, the scan continues to the right, marking the column where the first pixel value becomes non-zero as the right boundary of the first character. This process segments one character. The scan continues to the right, repeating the above operation until all columns have been scanned, thus completing the character segmentation for the current row in the binary image.

[0056] The segmented character image blocks may be single-character images or concatenated character image blocks. Whether a character is concatenated can be determined later by the width of the obtained character image.

[0057] S600, Confirm the category of the first character image block.

[0058] The first character image block is a single-character image block, as it contains only one character, and its width is basically the same as the width of a single character.

[0059] The first character image block is a connected character image block, because it is guaranteed that there are at least two connected characters, and its width is at least the width of two characters.

[0060] Based on this consideration, the categories of all first character image blocks can be identified one by one in the following manner:

[0061] The width of the first character image block is compared with a predetermined threshold to determine the category of the first character image block. Specifically, when the width of the first character image block is less than the predetermined threshold, the first character image block is determined as a single character image block; otherwise, the first character image block is determined as a connected character image block.

[0062] The predetermined threshold is slightly larger than the width of a single character.

[0063] S700. If the first character image block is a single character image block, then the single character in the single character image block is recognized by matching it with the template character image.

[0064] For example, the template character image includes at least 26 uppercase English letters and 26 lowercase English letters. To enable precise matching, optionally, the first character image block and the template character image need to be resized uniformly before matching, ensuring that their sizes are identical.

[0065] S800. If the first character image block is a contiguous character image block, then the contiguous character image block is de-contiguously processed to recognize each character in the contiguous character image block.

[0066] In the prior art, there are already a large number of mature de-adhesion processing methods. Step S800 of the present invention can directly adopt these mature de-adhesion processing methods to divide the adhering character image block into multiple single-character image blocks that each contain only one single character, and then identify the characters in the single-character image blocks one by one.

[0067] However, these methods for removing adhering characters are generally complex and require significant computational resources. Therefore, this invention provides a preferred method for removing adhering character image blocks, the specific process of which is as follows:

[0068] like Figure 2 As shown, step S800 includes the following sub-steps:

[0069] S810: Perform connected component analysis on the connected character image block, and coarsely divide the connected character image block into several second image character blocks according to the high and low positions in the connected component analysis results. The second image character block is an uppercase single character graphic block including an uppercase character, or a lowercase single character graphic block including a lowercase single character, or an equal-height connected character image block including at least a number of uppercase single characters or a number of lowercase single characters.

[0070] In a patch of connected characters, although adjacent characters are connected, the height of uppercase characters is significantly greater than that of lowercase characters. Therefore, during connected component analysis of the patch, if two consecutive regions have significantly different heights, these two regions can be segmented. Ultimately, the patch of connected characters can be divided into several second image character patches.

[0071] Of course, there is also an extreme case where all characters in the contiguous character image block are either all uppercase characters or all lowercase characters with the same height. In this case, coarse segmentation cannot be performed; that is, the second image character block output by S810 is actually the original, unprocessed contiguous character image block.

[0072] S820: Confirm the category of the second character image block.

[0073] The specific confirmation process is the same as step S600 in the previous text, specifically as follows:

[0074] The width of the second character image block is compared with a predetermined threshold to determine the category of the second character image block:

[0075] When the width of the second character image block is less than a predetermined threshold, the second character image block is determined as either an uppercase single character graphic block or a lowercase single character graphic block.

[0076] Otherwise, the first character image block is determined as a character image block with equal height and overlapping. Note that the difference from step S600 is that the character image block with equal height and overlapping at this time includes multiple characters with overlapping that are all uppercase characters of the same height, or all lowercase characters.

[0077] S830: If the second character image block is an uppercase single character graphic block or a lowercase single character graphic block, then by matching it with the template character image, the uppercase single character or lowercase single character in the uppercase single character graphic block or lowercase single character graphic block is recognized.

[0078] The specific matching process is the same as step S700 in the previous text, and will not be repeated here.

[0079] S840: If the second character image block is a character image block with equal height and overlapping, then further de-coupling processing is performed on the character image block with equal height and overlapping to realize the recognition of each character in the character image block with equal height and overlapping.

[0080] In the specific implementation process, various existing de-adhesion processing methods can be used to further de-adhesion process the image blocks of characters with equal height and adhesion, so as to extract each character and recognize them one by one.

[0081] This invention also provides a preferred method for de-adhesion processing of image blocks with characters of equal height that are stuck together. The specific process is as follows:

[0082] like Figure 3 As shown, step S840 may further include the following sub-steps:

[0083] S841: Resize the image T of characters with equal height and adhesion to obtain the processed image T of characters with equal height and adhesion to the target size. new .

[0084] S842: Based on proportion T new The template character image M is resized using _h / M_h to obtain the processed character image T that is contiguous with the same height. new Template character images M with the same height new ,in:

[0085] T new _h represents the processed image of the contour-aligned characters T. new The height value, M_h is the height value of the template character image before processing;

[0086] S843: Template character image M after processing based on template character image new Width dimensions of the image T of the character with equal height adhesion new The image is cropped to extract the third character image containing a single character.

[0087] S843: Match the extracted third character image template character image to recognize single characters in the third character image.

[0088] This completes the recognition of all characters in the image displayed on the control system screen, allowing for accurate understanding of the patient's treatment plan.

[0089] To enable those skilled in the art to more clearly understand the technical solution of the present invention, the following will be combined with Figure 4 and Figure 5 This paper introduces the implementation process of the present invention from another perspective through a more specific identification case.

[0090] like Figures 4 to 5 As shown, the specific implementation process of this identification case is as follows:

[0091] Step 1: Acquire the image from the treatment control system display.

[0092] Step 2: Obtain the region to be identified from the image, which is the patient planning information region. This patient planning information region is the image to be identified.

[0093] Step 3: Perform binarization processing on the image to be recognized to obtain the following result: Figure 5 (a) shows the binarized image.

[0094] Step 4: Segment the binarized image horizontally and vertically to obtain the bounding rectangle containing the target character. To achieve this... Figure 5 As shown in (b), the rectangle added to the image represents the circumscribed rectangle.

[0095] Step 5: Perform character segmentation on the binarized image from left to right to obtain several characters. Determine if the width of each character is greater than or equal to a width threshold A to determine if the character is a connected character (T). Specifically, if the character width is less than A, it is considered a single character (S). The template characters M and S are resized to maintain the same size. A matching algorithm is used to calculate the matching rate between the two characters, and the character with the highest matching rate is the final recognition result. If the width is greater than A, it is considered a connected character T, such as... Figure 5 As shown in (c).

[0096] Step 6: Perform connectivity analysis on the connected character T, and coarsely segment it into uppercase and lowercase single characters or connected characters based on the high and low positions of the analysis results, such as... Figure 5 As shown in (d).

[0097] Step 7: Divide the above results using left-right and top-bottom segmentation methods. The segmentation results are as follows: Figure 5 As shown in (e) and 5(f).

[0098] Step 8: Sequentially determine whether the character widths in 5(e) and 5(f) are greater than the threshold A. If they are less than A, it indicates a single character (S). Resize the template character (M) and S to maintain the same size, and calculate their matching rate using a matching algorithm. The character with the highest matching rate is the final recognition result. If they are greater than A, it indicates a concatenated character T. First, resize the concatenated character T to obtain T. new Then, based on the proportion (T) new Resize M by (_h / M_h) to maintain the same height, resulting in M. new Then, based on M... new Width Mnew _w to T new Perform trimming to obtain T1 and T2, as follows: Figure 5 As shown in (g), the left side represents T1 and the right side represents T2.

[0099] Step 9: Calculate the matching rate between T1 and M, and obtain the character with the highest matching rate, which is the final recognition result. Re-evaluate the width of T2 and repeat the above operation until all characters have been recognized.

[0100] The present invention also provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the character recognition method in the display of the treatment control system provided by the present invention.

[0101] The present invention also provides an electronic device, wherein a computer program is stored on the computer-readable storage medium, and when the program is executed by a processor, it implements the character recognition method in the display of the treatment control system provided by the present invention.

[0102] The present invention has been described above in sufficient detail and with certain specificities. Those skilled in the art should understand that the descriptions in the embodiments are merely exemplary, and all changes made without departing from the true spirit and scope of the invention should fall within the protection scope of the invention. The scope of protection claimed by the present invention is defined by the claims, and not by the above descriptions in the embodiments.

Claims

1. A character recognition method in a display of a treatment control system, characterized in that, It includes: Acquire images from the display of the treatment control system; Obtain an image to be recognized, including characters, from a predetermined region of the image; The image to be recognized is binarized to obtain a binary image; Remove the blank areas at the edges of the binarized image; Character segmentation is performed on the binarized image after removing edge blank areas to obtain several first character image blocks, where the first character image blocks are single character image blocks or connected character image blocks; The category of the first character image block is determined, and the following processing is performed: If the first character image block is a single character image block, then the single character in the single character image block is identified by matching it with the template character image; If the first character image block is a connected character image block, then the connected character image block is subjected to de-connection processing in order to recognize each character in the connected character image block; The step of performing de-adhesion processing on the image blocks of adhered characters to recognize each character in the image blocks includes: Connectivity analysis is performed on the connected character image block. Based on the high and low positions in the connected component analysis results, the connected character image block is coarsely divided into several second character image blocks. The second character image block is an uppercase single character graphic block including an uppercase character, or a lowercase single character graphic block including a lowercase single character, or an equal-height connected character image block including at least a number of uppercase single characters or a number of lowercase single characters. Confirm the category of the second character image block and perform the following processing: If the second character image block is an uppercase single character graphic block or a lowercase single character graphic block, then by matching it with the template character image, the uppercase single character or lowercase single character in the uppercase single character graphic block or lowercase single character graphic block is identified. If the second character image block is a character image block with equal height and sticking together, then further de-sticking processing is performed on the character image block with equal height and sticking together, so as to realize the recognition of each character in the character image block with equal height and sticking together. The process of confirming the category of the second character image block includes: The width of the second character image block is compared with a predetermined threshold to determine the category of the second character image block, specifically: If the width of the second character image block is less than the predetermined threshold, the second character image block is determined as an uppercase single character graphic block or a lowercase single character graphic block; otherwise, the second character image block is determined as a character image block with equal height and adhering to each other.

2. The character recognition method as described in claim 1, characterized in that, The binarized image, after removing edge blank areas, undergoes character segmentation to obtain several first character image blocks, including: The binarized image is segmented from left to right using connected component analysis. Remove the blank areas at the edges of each segmented image obtained from the segmentation to obtain several first character image blocks.

3. The character recognition method as described in claim 1, characterized in that, The category of the first character image block is confirmed as follows: The width of the first character image block is compared with a predetermined threshold to determine the category of the first character image block, specifically: If the width of the first character image block is less than the predetermined threshold, the first character image block is determined to be a single character image block; otherwise, the first character image block is determined to be a connected character image block.

4. The character recognition method as described in claim 1, characterized in that, The step of performing de-cohesion processing on the image blocks of characters with equal height to recognize each character in the image blocks of characters with equal height includes: The image of characters with equal height and overlapping edges is resized to obtain a processed image of characters with equal height and overlapping edges of the target size. ; According to proportion The template character image is resized using _h / M_h to obtain the processed character image that is connected to the same height. Template character images with the same height ,in, _h represents the processed image of the contour-aligned characters. The height value, M_h is the height value of the template character image before processing; Template character image after processing based on template character image Width dimensions of the image with equal height for the connected characters Perform cropping to extract image blocks containing a single third character one by one; The recognition of a single character in the third character image block is achieved by matching it with a template character image.

5. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the character recognition method according to any one of claims 1 to 4.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the character recognition method according to any one of claims 1 to 4.

Images