Electronic laryngoscope image recognition method and system for children's airway foreign bodies

Abstract

The present invention relates to the field of image processing technology, and in particular to an electronic laryngoscope image recognition method and system for children's airway foreign bodies, comprising: dividing an electronic laryngoscope image into a suspected fishbone area and an airway tissue area, obtaining a reference airway tissue area of ​​any suspected fishbone area, analyzing the shape characteristics of the suspected fishbone area and the reference airway tissue area, selecting a positioning point and a reference point, obtaining a positioning index of the suspected fishbone area, analyzing the fishbone possibility of the positioning index and the fishbone possibility of the shape of the suspected fishbone area of ​​different electronic laryngoscope images, and determining the real fishbone area. The present invention can effectively avoid the change of the position of the fishbone in the image caused by the change of the laryngoscope position and angle, can effectively avoid the interference of the reflective part on the airway tissue on the recognition of the fishbone foreign body, and improve the accuracy of the recognition of the children's airway foreign body.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to an electronic laryngoscope image recognition method and system for foreign bodies in children's airways. Background Art

[0002] Electronic laryngoscope images are images of the inside of the airway taken by an electronic laryngoscope. Electronic laryngoscopes use built-in cameras and light sources to transmit the situation in the airway to the display in real time, allowing doctors to directly observe the location, size and shape of foreign bodies, which is crucial for rapid and accurate diagnosis and location of foreign bodies. Fishbones are common foreign bodies in children's airway foreign bodies. When a child is stuck with a fishbone, if it is not discovered and removed in time, it will not only damage the child's airway, but may even be life-threatening.

[0003] The traditional method of identifying fishbones in the airway is for doctors to directly observe and identify fishbones in the airway by observing the laryngoscope image in real time. However, children's airways are narrower, and the image stability cannot always be maintained during the laryngoscopy. In addition, due to lighting, the inner wall of the tissue will produce reflections, which have a certain similarity to fishbones. These factors increase the difficulty for doctors to identify fishbones in children's airways, interfere with doctors' timely and accurate detection of fishbones, and increase the harm of fishbones to children. Summary of the invention

[0004] The present invention provides an electronic laryngoscope image recognition method and system for foreign bodies in children's airways, so as to solve the existing problems.

[0005] The electronic laryngoscope image recognition method and system for children's airway foreign bodies of the present invention adopts the following technical solutions:

[0006] One embodiment of the present invention provides an electronic laryngoscope image recognition method for foreign bodies in children's airways, the method comprising the following steps:

[0007] Acquire several electronic laryngoscope images;

[0008] Perform image segmentation on any electronic laryngoscope image to obtain a number of regions to be analyzed; divide the regions to be analyzed into suspected fishbone regions and airway tissue regions according to the color expression of any regions to be analyzed;

[0009] According to the positional relationship between the suspected fishbone area and the adjacent airway tissue area, and the edge similarity between the suspected fishbone area and the adjacent airway tissue area, a reference airway tissue area is obtained for any suspected fishbone area; shape features of the suspected fishbone area and the reference airway tissue area are analyzed, and a positioning point of the suspected fishbone area and a reference point corresponding to the reference airway tissue area are selected; and a positioning index of the suspected fishbone area is obtained according to the distance relationship between the positioning point and the reference point;

[0010] The fishbone possibility of the suspected fishbone area in different electronic laryngoscope images with respect to the positioning index and the fishbone possibility with respect to the shape are analyzed to determine the real fishbone area in any electronic laryngoscope image.

[0011] Furthermore, the image segmentation of any electronic laryngoscope image to obtain a plurality of regions to be analyzed includes the following specific methods:

[0012] For any electronic laryngoscope image, the pixel with the largest sum of the R, G, and B channel values ​​is used as the seed point, and the difference of the R, G, and B channel values ​​is within A as the condition for the seed point to grow. When there are no pixels that meet the growth conditions, the growth is stopped, and the growing area is used as an area to be analyzed; after the growth is stopped, the pixel with the largest sum of the R, G, and B channel values ​​in the pixel points not in the area to be analyzed is used as the seed point, and the area growth is continued; and so on, until all the pixels have their own area to be analyzed;

[0013] The A mentioned above are all preset values.

[0014] Furthermore, the method of dividing the area to be analyzed into a suspected fishbone area and an airway tissue area according to the color expression of any area to be analyzed includes:

[0015] Get the probability that any area to be analyzed is a suspected fishbone area:

[0016]

[0017] in, Indicates the possibility that any area to be analyzed is a suspected fishbone area. Represents the number of pixels in the area to be analyzed; , , Indicates the first The R, G, and B channel values ​​of each pixel; represents the normalization function;

[0018] Among all the areas to be analyzed in any electronic laryngoscope image, the areas to be analyzed whose possibility of being suspected fishbone areas is greater than or equal to a preset first parameter are recorded as suspected fishbone areas of the electronic laryngoscope image, and the areas to be analyzed that are not suspected fishbone areas are recorded as airway tissue areas of the electronic laryngoscope image.

[0019] Furthermore, the method of obtaining a reference airway tissue region of any suspected fishbone region according to the positional relationship between the suspected fishbone region and the adjacent airway tissue region and the edge similarity between the suspected fishbone region and the adjacent airway tissue region includes the following specific methods:

[0020] When the suspected fishbone area is surrounded by an airway tissue area, the airway tissue area is recorded as a reference airway tissue area of ​​the suspected fishbone area;

[0021] When the suspected fishbone region is connected to several airway tissue regions, obtain H pixel points on the edge of any connected airway tissue region that are adjacent to the edge of the connected portion, calculate the gradient sum of the H pixel points, and subtract the gradient sum of the pixel points on the edge of the connected portion and take the absolute value to obtain the edge gradient difference between the airway tissue region and the suspected fishbone region, and record the airway tissue region with the smallest edge gradient difference as the reference airway tissue region of the suspected fishbone region;

[0022] The H is a preset second parameter.

[0023] Furthermore, the analyzing the shape features of the suspected fishbone region and the reference airway tissue region, and selecting the positioning point of the suspected fishbone region and the reference point of the corresponding reference airway tissue region, includes the following specific methods:

[0024] The two pixels with the largest distance in any suspected fishbone area are used as the positioning points of the suspected fishbone area;

[0025] The two pixel points with the largest distance on the edge of any reference airway tissue region are used as reference points of the reference airway tissue region.

[0026] Furthermore, the method of obtaining the location index of the suspected fishbone area according to the distance relationship between the location point and the reference point includes the following specific methods:

[0027] Get the distance between all positioning points and all reference points;

[0028] The distance between any positioning point and any reference point is compared with the distance between two reference points to obtain the positioning index of the positioning point and the reference point of the corresponding reference airway tissue area.

[0029] Furthermore, the fishbone possibility of the suspected fishbone areas in different electronic laryngoscope images with respect to the positioning index and the fishbone possibility with respect to the shape are analyzed to determine the real fishbone area in any electronic laryngoscope image, including the specific method of:

[0030] Reference airway tissue regions were matched;

[0031] Obtaining the fishbone possibility of the suspected fishbone region corresponding to the matched reference airway tissue region with respect to the positioning index and the fishbone possibility with respect to the shape;

[0032] According to the fishbone possibility of the suspected fishbone region with respect to the positioning point and the fishbone possibility of the suspected fishbone region with respect to the shape, the possibility of any fishbone region being a real fishbone region is obtained, wherein the fishbone possibility of the suspected fishbone region with respect to the positioning point and the fishbone possibility of the suspected fishbone region with respect to the shape are both in direct proportion to the possibility of the fishbone region being a real fishbone region;

[0033] The suspected fishbone region whose possibility of being a real fishbone region is greater than a preset third parameter is recorded as a real fishbone region.

[0034] Furthermore, the specific method for obtaining the matching reference airway tissue region is:

[0035] A boundary tracking algorithm is used to obtain the chain code of the edge of the reference airway tissue area in any electronic laryngoscope image, and the edit distance of the chain code of any reference airway tissue area edge in different electronic laryngoscope images is obtained. The two airway tissue area edges with the smallest edit distance are the same reference airway tissue area.

[0036] Furthermore, the specific method for obtaining the fishbone possibility of the suspected fishbone region corresponding to the matched reference airway tissue region with respect to the positioning point and the fishbone possibility with respect to the shape is:

[0037] For two suspected fishbone areas corresponding to the reference airway tissue areas matched in any two electronic laryngoscope images, the absolute value of the difference of the positioning index of each positioning point of the two suspected fishbone areas is obtained, and the difference of the positioning points of the two suspected fishbone areas is obtained by summing them up; if the positioning point difference is less than or equal to the preset fourth parameter, and the edges of the reference airway tissue areas of the two suspected fishbone areas are the same, then the positioning points are matched, and the two suspected fishbone areas are matched; the number of matches of any suspected fishbone area in all electronic laryngoscope images is compared with the number of electronic laryngoscope images to obtain the fishbone possibility of the suspected fishbone area with respect to the positioning point;

[0038] For two suspected fishbone areas corresponding to the matched reference airway tissue areas in any two electronic laryngoscope images, a boundary tracking algorithm is used to obtain the edge chain codes of the two suspected fishbone areas, and the edit distance of the edge chain codes of the two suspected fishbone areas is calculated; if the edit distance is less than or equal to the preset fifth parameter, the chain codes of the two suspected fishbone areas match; the number of times the chain code matching of any suspected fishbone area appears in all electronic laryngoscope images is compared with the number of electronic laryngoscope images to obtain the fishbone possibility of the suspected fishbone area with respect to the shape.

[0039] An embodiment of the present invention provides an electronic laryngoscope image recognition system for foreign bodies in children's airways, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any one of the steps of the electronic laryngoscope image recognition method for foreign bodies in children's airways.

[0040] The beneficial effects of the technical solution of the present invention are: using the adjacent airway tissue area of ​​the suspected fishbone area as a positioning reference for the suspected fishbone area, comparing the similarity of the position of the suspected fishbone area in different frames of laryngoscopic images, and then determining the same suspected fishbone area in different laryngoscopic images, can effectively avoid the problem that the position of the fishbone in the image changes due to changes in the position and angle of the laryngoscope, and cannot be directly matched. According to the characteristics that the shape and position of the airway tissue reflection will change under different lighting and shooting angles, while the fishbone will not change, the shape and position of the suspected fishbone area in the continuous laryngoscopic images are compared to determine the real fishbone area, which can effectively avoid the interference of the reflective part on the airway tissue on the identification of the fishbone foreign body. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0042] Figure 1 The present invention is a flowchart of the steps of the electronic laryngoscope image recognition method for foreign bodies in children's airways.

[0043] Figure 2 It is a fishbone schematic diagram of the electronic laryngoscope image recognition method for foreign bodies in children's airways according to the present invention.

[0044] Figure 3 It is a schematic diagram of the change of the laryngoscope reflective shape of the electronic laryngoscope image recognition method for children's airway foreign bodies of the present invention. DETAILED DESCRIPTION

[0045] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the electronic laryngoscope image recognition method and system for children's airway foreign bodies proposed by the present invention are described in detail below in combination with the accompanying drawings and preferred embodiments, and its specific implementation, structure, features and effects are described as follows. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0046] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0047] The specific scheme of the electronic laryngoscope image recognition method and system for children's airway foreign bodies provided by the present invention is described in detail below with reference to the accompanying drawings.

[0048] See also Figure 1 , which shows a flowchart of the steps of an electronic laryngoscope image recognition method for children's airway foreign bodies provided by an embodiment of the present invention, the method comprising the following steps:

[0049] Step S001: Acquire several electronic laryngoscope images.

[0050] Specifically, in order to implement the electronic laryngoscope image recognition method for children's airway foreign bodies proposed in this embodiment, it is first necessary to collect electronic laryngoscope images of children's airways. The specific process is as follows:

[0051] The video collected by the laryngoscope during the laryngoscopy of the child is obtained, the collection cycle is preset, and C electronic laryngoscope images are captured in the video.

[0052] It should be noted that C is a preset value. According to experience, C is set to 10 and the sampling period is set to 3 frames. It can be adjusted according to actual conditions and is not specifically limited in this embodiment.

[0053] So far, several electronic laryngoscope images are obtained through the above method.

[0054] Step S002: performing image segmentation on any electronic laryngoscope image to obtain a plurality of regions to be analyzed; dividing the regions to be analyzed into suspected fishbone regions and airway tissue regions according to the color expression of any regions to be analyzed.

[0055] It should be noted that since both the fishbone and the reflection of the inner wall of the tissue have large three-channel values, and when the fishbone penetrates deeply into the inner wall of the airway, the fishbone and the reflection area are relatively small, the suspected fishbone area in the electronic laryngoscope image can be obtained based on the color characteristics of the fishbone in the electronic laryngoscope image.

[0056] Schematic diagram of a fish bone in the airway Figure 2 shown.

[0057] Specifically, in step 2.1, image segmentation is performed on any electronic laryngoscope image to obtain a plurality of regions to be analyzed.

[0058] This embodiment uses a region growing algorithm to segment any electronic laryngoscope image. The specific segmentation process is as follows:

[0059] For any electronic laryngoscope image, the pixel with the largest sum of the R, G, and B channel values ​​is used as the seed point, and the difference of the R, G, and B channel values ​​is within A as the condition for the seed point to grow. When there are no pixels that meet the growth conditions, the growth is stopped and the growing area is used as an area to be analyzed. After the growth is stopped, the pixel with the largest sum of the R, G, and B channel values ​​in the pixel points not in the area to be analyzed is used as the seed point to continue the regional growth; and so on, until all pixels have their own area to be analyzed.

[0060] It should be noted that A is a preset value, and A is set to 2 based on experience. It can be adjusted according to actual conditions, and this embodiment does not specifically limit it.

[0061] Step 2.2, according to the color expression of any area to be analyzed, the area to be analyzed is divided into a suspected fishbone area and an airway tissue area.

[0062] It should be noted that since fish bones appear white or transparent (but have low light transmittance), the three-channel values ​​of the pixels in the fish bone area should be larger.

[0063] Therefore, the possibility of any area to be analyzed being a suspected fishbone area is calculated based on the mean of the R, G, and B channel values ​​of the pixels in the area to be analyzed.

[0064] As an embodiment, the possibility of any area to be analyzed being a suspected fishbone area is obtained:

[0065]

[0066] in, Indicates the possibility that any area to be analyzed is a suspected fishbone area. Represents the number of pixels in the area to be analyzed; , , Indicates the first The R, G, and B channel values ​​of each pixel; Represents the normalization function.

[0067] It should be noted that It represents the average value of the three-channel values ​​of the pixels in the area to be analyzed. The larger the value is, the more likely it is a fishbone area.

[0068] Among all the areas to be analyzed in any electronic laryngoscope image, the areas to be analyzed whose possibility of being suspected fishbone areas is greater than or equal to a preset first parameter are recorded as suspected fishbone areas of the electronic laryngoscope image, and the areas to be analyzed that are not suspected fishbone areas are recorded as airway tissue areas of the electronic laryngoscope image.

[0069] It should be noted that the preset first parameter is set to 0.7 based on experience, and can be adjusted according to actual conditions, and is not specifically limited in this embodiment.

[0070] So far, the suspected fishbone region of any electronic laryngoscope image is obtained by the above method.

[0071] Step S003: Obtain a reference airway tissue region of any suspected fishbone region based on the positional relationship between the suspected fishbone region and the adjacent airway tissue region, and the edge similarity between the suspected fishbone region and the adjacent airway tissue region; analyze the shape features of the suspected fishbone region and the reference airway tissue region, select the positioning point of the suspected fishbone region and the reference point of the corresponding reference airway tissue region; obtain the positioning index of the suspected fishbone region based on the distance relationship between the positioning point and the reference point.

[0072] It should be noted that, since the image will not remain stable when the laryngoscope goes deep into the airway, in order to match the suspected fishbone area in the laryngoscope images of different frames, it is necessary to find a position that can be used as a reference to locate the suspected fishbone area. Since the edge of the airway tissue area is relatively clear and will not change, the edge of the airway tissue area is used as the reference position.

[0073] Specifically, in step 3.1, a reference airway tissue region of any suspected fishbone region is obtained according to the positional relationship between the suspected fishbone region and the adjacent airway tissue region and the edge similarity between the suspected fishbone region and the adjacent airway tissue region.

[0074] It should be noted that, step S002 divides any electronic laryngoscope image into a suspected fishbone area and an airway tissue area. Therefore, there are two types of adjacent relationships between the airway tissue areas adjacent to the suspected fishbone area. One is that the suspected fishbone area is completely surrounded by an airway tissue area and can be used directly for reference; the other is that the suspected fishbone area has several adjacent airway tissue areas, and one should be selected for reference. At this time, the one that is more similar to the fishbone shape should be selected, that is, the connected edges extending out should be as similar as possible.

[0075] When the suspected fishbone area is surrounded by an airway tissue area, the airway tissue area is recorded as a reference airway tissue area of ​​the suspected fishbone area;

[0076] When the suspected fishbone region is connected to several airway tissue regions, obtain H pixel points on the edge of any connected airway tissue region that are adjacent to the edge of the connected portion, calculate the gradient sum of the H pixel points, and subtract the gradient sum of the pixel points on the edge of the connected portion and take the absolute value to obtain the edge gradient difference between the airway tissue region and the suspected fishbone region, and record the airway tissue region with the smallest edge gradient difference as the reference airway tissue region of the suspected fishbone region;

[0077] The H is a preset second parameter.

[0078] It should be noted that the preset second parameter H is set to 20 based on experience, and can be adjusted according to actual conditions, and is not specifically limited in this embodiment.

[0079] Step 3.2, analyzing the shape features of the suspected fishbone area and the reference airway tissue area, and selecting the positioning point of the suspected fishbone area and the reference point corresponding to the reference airway tissue area.

[0080] It should be noted that, since fish bones are usually long and thin, the pixels at the two ends of the longest one can be used for positioning.

[0081] The two pixels with the largest distance in any suspected fishbone area are used as the positioning points of the suspected fishbone area.

[0082] It should be noted that, because the shape of the tissue region does not change, the two edge pixel points that are farthest apart can be found in any reference airway tissue region for reference.

[0083] The two pixel points with the largest distance on the edge of any reference airway tissue region are used as reference points of the reference airway tissue region.

[0084] Step 3.3, based on the distance relationship between the positioning point and the reference point, obtain the positioning index of the suspected fishbone area.

[0085] It should be noted that, during laryngoscopy, as the position of the laryngoscope moves, the fishbone and airway tissue may be enlarged or reduced in the image, so it is necessary to obtain the relative position of the positioning point and the reference point.

[0086] As an example, The method for obtaining the positioning index of a positioning point is:

[0087] Get the distance between all positioning points and all reference points;

[0088]

[0089] in, Indicates The first anchor point and the Positioning indicators of reference points; Indicates The first anchor point and the The distance between the reference points; Represents the distance between two reference points.

[0090] It should be noted that It represents the ratio of the spacing between the positioning point and the reference point to the spacing between the reference points, reflecting the relative distance of the positioning point relative to the constant spacing of the reference points. For the same suspected fishbone area on different electronic laryngoscope images, the positioning indicators of the positioning points should be similar.

[0091] At this point, the positioning index of the suspected fishbone area is obtained through the above method.

[0092] Step S004: analyzing the fishbone possibility of the suspected fishbone regions in different electronic laryngoscope images with respect to the positioning index and the fishbone possibility with respect to the shape, and determining the real fishbone region in any electronic laryngoscope image.

[0093] It should be noted that the position and shape of the reflection of the inner wall of the airway tissue relative to the edge of the tissue will change with the position and angle of the laryngoscope, so the reflection of the inner wall of the tissue in the multi-frame image of the video cannot match. However, when the fishbone is stuck inside the airway, its position relative to the airway tissue will not change when the position and angle of the laryngoscope change, and the shape of the fishbone will not change, so the same suspected fishbone area features will be shown in the multi-frame image.

[0094] Specifically, first, the reference airway tissue region is matched.

[0095] It should be noted that the airway tissue region hardly changes in shape during the detection process, that is, the coordinates of the edge of the airway tissue region remain unchanged, and matching can be performed through the difference between the chain codes at the edge of the airway tissue region.

[0096] A boundary tracking algorithm is used to obtain the chain code of the edge of the reference airway tissue area in any electronic laryngoscope image, and the edit distance of the chain code of any reference airway tissue area edge in different electronic laryngoscope images is obtained. The two airway tissue area edges with the smallest edit distance are the same reference airway tissue area.

[0097] It should be noted that the chain code is used to represent curves and region boundaries, and the edit distance is used to represent the degree of similarity between character strings. Both are existing methods, and therefore will not be described in detail in this embodiment.

[0098] It should be noted that the shape of the reflective area has a certain degree of randomness. Even if there is reflection at the same position, the shape of the reflection will change with the change of lighting or shooting angle; while the shape of fish bones will not change. Therefore, after matching the positioning points of the suspected fish bone area in different electronic laryngoscope images, its shape needs to be matched.

[0099] Schematic diagram of the change in the shape of the laryngoscope reflection Figure 3 As shown, , Electronic laryngoscope images at different angles.

[0100] Secondly, the fishbone possibility of the suspected fishbone area corresponding to the matched reference airway tissue area with respect to the positioning index and the fishbone possibility with respect to the shape are obtained.

[0101] For two suspected fishbone areas corresponding to the matched reference airway tissue areas in any two electronic laryngoscope images, the absolute value of the difference in positioning indicators of each positioning point of the two suspected fishbone areas is obtained, and the difference in positioning points of the two suspected fishbone areas is obtained; if the positioning point difference is less than or equal to the preset fourth parameter, and the edges of the reference airway tissue areas of the two suspected fishbone areas are the same, then the positioning points are matched, and the two suspected fishbone areas are matched; the number of matches of any suspected fishbone area in all electronic laryngoscope images is compared with the number of electronic laryngoscope images to obtain the fishbone possibility of the suspected fishbone area with respect to the positioning point.

[0102] It should be noted that the fourth parameter is preset to 0.2 based on experience and can be adjusted according to actual conditions, and is not specifically limited in this embodiment.

[0103] For two suspected fishbone areas corresponding to the matched reference airway tissue areas in any two electronic laryngoscope images, a boundary tracking algorithm is used to obtain the edge chain codes of the two suspected fishbone areas, and the edit distance of the edge chain codes of the two suspected fishbone areas is calculated; if the edit distance is less than or equal to the preset fifth parameter, the chain codes of the two suspected fishbone areas match; the number of times the chain code matching of any suspected fishbone area appears in all electronic laryngoscope images is compared with the number of electronic laryngoscope images to obtain the fishbone possibility of the suspected fishbone area with respect to the shape.

[0104] It should be noted that the preset fifth parameter is set to 40 based on experience, and can be adjusted according to actual conditions, and this embodiment does not make a specific limitation.

[0105] Then, according to the fishbone possibility of the suspected fishbone region with respect to the positioning point and the fishbone possibility of the suspected fishbone region with respect to the shape, the possibility of any fishbone region being a real fishbone region is obtained.

[0106] As an embodiment, the possibility that any suspected fishbone region is a real fishbone region is obtained as follows:

[0107]

[0108] in, Indicates the possibility that any suspected fishbone area is a real fishbone area, Indicates the possibility that the suspected fishbone area is the real fishbone area in the positioning point matching, It indicates the possibility that the suspected fishbone area is a real fishbone area in shape.

[0109] Finally, the suspected fishbone region whose possibility of being a real fishbone region is greater than the preset third parameter is recorded as a real fishbone region.

[0110] It should be noted that the preset third parameter is set to 0.7 based on experience, and can be adjusted according to actual conditions, and is not specifically limited in this embodiment.

[0111] Through the above steps, the electronic laryngoscope image recognition of foreign bodies in children's airway is completed.

[0112] An embodiment of the present invention provides an electronic laryngoscope image recognition system for foreign bodies in the airway of children. The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any one of the steps of the electronic laryngoscope image recognition method for foreign bodies in the airway of children.

[0113] This embodiment uses the adjacent airway tissue area of ​​the suspected fishbone area as a positioning reference for the suspected fishbone area, compares the similarity of the position of the suspected fishbone area in different frames of laryngoscopic images, and then determines the same suspected fishbone area in different laryngoscopic images, which can effectively avoid the problem that the position of the fishbone in the image changes due to changes in the position and angle of the laryngoscope, and cannot be directly matched. According to the characteristics that the shape and position of the airway tissue reflection will change under different lighting and shooting angles, while the fishbone will not change, the shape and position of the suspected fishbone area in the continuous laryngoscopic images are compared to determine the real fishbone area, which can effectively avoid the interference of the reflective part on the airway tissue on the identification of the fishbone foreign body.

[0114] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An electronic laryngoscope image recognition method for foreign bodies in children's airways, characterized in that: The method comprises the following steps: Acquire several electronic laryngoscope images; Perform image segmentation on any electronic laryngoscope image to obtain a number of regions to be analyzed; divide the region to be analyzed into suspected fishbone regions and airway tissue regions according to the color expression of any region to be analyzed; According to the positional relationship between the suspected fishbone area and the adjacent airway tissue area, and the edge similarity between the suspected fishbone area and the adjacent airway tissue area, a reference airway tissue area is obtained for any suspected fishbone area; shape features of the suspected fishbone area and the reference airway tissue area are analyzed, and a positioning point of the suspected fishbone area and a reference point corresponding to the reference airway tissue area are selected; and a positioning index of the suspected fishbone area is obtained according to the distance relationship between the positioning point and the reference point; Analyze the fishbone possibility of the suspected fishbone area in different electronic laryngoscope images with respect to the positioning index and the fishbone possibility with respect to the shape, and determine the real fishbone area in any electronic laryngoscope image; The method of obtaining a reference airway tissue region of any suspected fishbone region according to the positional relationship between the suspected fishbone region and the adjacent airway tissue region and the edge similarity between the suspected fishbone region and the adjacent airway tissue region includes: When the suspected fishbone area is surrounded by an airway tissue area, the airway tissue area is recorded as a reference airway tissue area of ​​the suspected fishbone area; When the suspected fishbone region is connected to several airway tissue regions, obtain H pixel points on the edge of any connected airway tissue region that are adjacent to the edge of the connected portion, calculate the gradient sum of the H pixel points, and subtract the gradient sum of the pixel points on the edge of the connected portion and take the absolute value to obtain the edge gradient difference between the airway tissue region and the suspected fishbone region, and record the airway tissue region with the smallest edge gradient difference as the reference airway tissue region of the suspected fishbone region; The H is a preset second parameter; The specific method of analyzing the shape features of the suspected fishbone area and the reference airway tissue area and selecting the positioning point of the suspected fishbone area and the reference point corresponding to the reference airway tissue area includes: The two pixels with the largest distance in any suspected fishbone area are used as the positioning points of the suspected fishbone area; Taking the two pixel points with the largest distance on the edge of any reference airway tissue region as reference points of the reference airway tissue region; The method of analyzing the fishbone possibility of the positioning index and the fishbone possibility of the shape of the suspected fishbone areas in different electronic laryngoscope images to determine the real fishbone area in any electronic laryngoscope image includes the following specific methods: Reference airway tissue regions were matched; Obtaining the fishbone possibility of the suspected fishbone region corresponding to the matched reference airway tissue region with respect to the positioning index and the fishbone possibility with respect to the shape; According to the fishbone possibility of the suspected fishbone region with respect to the positioning point and the fishbone possibility of the suspected fishbone region with respect to the shape, the possibility of any fishbone region being a real fishbone region is obtained, wherein the fishbone possibility of the suspected fishbone region with respect to the positioning point and the fishbone possibility of the suspected fishbone region with respect to the shape are both in direct proportion to the possibility of the fishbone region being a real fishbone region; The suspected fishbone region whose possibility of being a real fishbone region is greater than a preset third parameter is recorded as a real fishbone region; The specific method for obtaining the fishbone possibility of the suspected fishbone area corresponding to the matched reference airway tissue area with respect to the positioning point and the fishbone possibility with respect to the shape is: For two suspected fishbone areas corresponding to the reference airway tissue areas matched in any two electronic laryngoscope images, the absolute value of the difference of the positioning index of each positioning point of the two suspected fishbone areas is obtained, and the difference of the positioning points of the two suspected fishbone areas is obtained by summing them up; if the positioning point difference is less than or equal to the preset fourth parameter, and the edges of the reference airway tissue areas of the two suspected fishbone areas are the same, then the positioning points are matched, and the two suspected fishbone areas are matched; the number of matches of any suspected fishbone area in all electronic laryngoscope images is compared with the number of electronic laryngoscope images to obtain the fishbone possibility of the suspected fishbone area with respect to the positioning point; For two suspected fishbone areas corresponding to the matched reference airway tissue areas in any two electronic laryngoscope images, a boundary tracking algorithm is used to obtain the edge chain codes of the two suspected fishbone areas, and the edit distance of the edge chain codes of the two suspected fishbone areas is calculated; if the edit distance is less than or equal to the preset fifth parameter, the chain codes of the two suspected fishbone areas match; the number of times the chain code matching of any suspected fishbone area appears in all electronic laryngoscope images is compared with the number of electronic laryngoscope images to obtain the fishbone possibility of the suspected fishbone area with respect to the shape.

2. The electronic laryngoscope image recognition method for children's airway foreign bodies according to claim 1, characterized in that: The specific method of performing image segmentation on any electronic laryngoscope image to obtain a plurality of regions to be analyzed includes: For any electronic laryngoscope image, the pixel with the largest sum of the R, G, and B channel values ​​is used as the seed point, and the difference of the R, G, and B channel values ​​is within A as the condition for the seed point to grow. When there are no pixels that meet the growth conditions, the growth is stopped, and the growing area is used as an area to be analyzed; after the growth is stopped, the pixel with the largest sum of the R, G, and B channel values ​​in the pixel points not in the area to be analyzed is used as the seed point, and the area growth is continued; and so on, until all the pixels have their own area to be analyzed; The A mentioned above are all preset values.

3. The electronic laryngoscope image recognition method for children's airway foreign bodies according to claim 1, characterized in that: The method of dividing the area to be analyzed into a suspected fishbone area and an airway tissue area according to the color expression of any area to be analyzed includes: Get the probability that any area to be analyzed is a suspected fishbone area: in, Indicates the possibility that any area to be analyzed is a suspected fishbone area. Represents the number of pixels in the area to be analyzed; , , Indicates the first The R, G, and B channel values ​​of each pixel; represents the normalization function; Among all the areas to be analyzed in any electronic laryngoscope image, the areas to be analyzed whose possibility of being suspected fishbone areas is greater than or equal to a preset first parameter are recorded as suspected fishbone areas of the electronic laryngoscope image, and the areas to be analyzed that are not suspected fishbone areas are recorded as airway tissue areas of the electronic laryngoscope image.

4. The electronic laryngoscope image recognition method for children's airway foreign bodies according to claim 1, characterized in that: The specific method of obtaining the positioning index of the suspected fishbone area according to the distance relationship between the positioning point and the reference point is as follows: Get the distance between all positioning points and all reference points; The distance between any positioning point and any reference point is compared with the distance between two reference points to obtain the positioning index of the positioning point and the reference point of the corresponding reference airway tissue area.

5. The electronic laryngoscope image recognition method for children's airway foreign bodies according to claim 1, characterized in that: The specific method for obtaining the matching reference airway tissue region is: A boundary tracking algorithm is used to obtain the chain code of the edge of the reference airway tissue area in any electronic laryngoscope image, and the edit distance of the chain code of any reference airway tissue area edge in different electronic laryngoscope images is obtained. The two airway tissue area edges with the smallest edit distance are the same reference airway tissue area.

6. An electronic laryngoscope image recognition system for foreign bodies in children's airways, 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 computer program, the steps of the electronic laryngoscope image recognition method for foreign bodies in the airway of children as described in any one of claims 1 to 5 are implemented.

Images