An engine design evaluation method based on machine vision

Abstract

The present application relates to an engine design evaluation method based on machine vision. The method includes: collecting images of different surfaces of the engine as an image set to be analyzed, where the image set to be analyzed includes at least the front image and the back image of the engine, preprocessing the image set to be analyzed to obtain a processed image set, performing image recognition and analysis on the processed image set to determine the density, integrity, cold and warm weighted value, color activity, regularity, and balance of the engine; analyzing the beauty degree of the engine using the beauty degree analysis formula according to the density, integrity, regularity, and balance of the engine; and outputting a design evaluation result according to the density, cold and warm weighted value, color activity, integrity, regularity, beauty degree, and balance. Thus, the engine design evaluation can be more standardized, and further improve the accuracy of the engine beauty degree evaluation.

Description

Technical Field

[0001] The present application relates to the field of computer vision technology, and particularly to an engine design evaluation method based on machine vision. Background Art

[0002] The design beauty degree of an engine is not a characteristic of the engine itself, but a concept in the fields of art, design, vision, and feeling, etc. However, considering the design beauty degree in the engine design process can help us build better, more beautiful, more user-friendly, and more suitable engines. And the design beauty degree is also a reference factor in the usual research and evaluation of engines. At present, the design of the engine is mainly subjectively judged by people, and there is no judgment standard, resulting in a low accuracy of engine design evaluation. Summary of the Invention

[0003] Based on this, in view of the above technical problems, it is necessary to provide an engine design evaluation method based on machine vision that can improve the accuracy of engine design evaluation.

[0004] An engine design evaluation method based on machine vision, the method comprising:

[0005] Collecting images of different surfaces of the engine as a set of images to be analyzed, the set of images to be analyzed including at least a front image and a back image of the engine;

[0006] Preprocessing the set of images to be analyzed to obtain a processed set of images;

[0007] Performing image recognition on the processed set of images to obtain the engine area of each image in the processed set of images, and the area of the dense regions of pipelines, lines, and components in each image, and determining the density of the engine by using a density analysis formula according to the engine area of each image and the area of each dense region;

[0008] Performing recognition on the front image or the back image in the processed set of images to obtain the rectangular area of the minimum circumscribed rectangle of the engine in the front image or the back image, and the engine area in the front image or the back image, and determining the integrity of the engine by using an integrity analysis formula according to the rectangular area and the engine area in the front image or the back image;

[0009] After dividing each image in the processed image set into grids, identify the HSV average value of each grid of each image and the engine area in each image. According to the HSV average value of each grid of each image, determine the color warmth and coldness of each grid of each image and the total area of the grids with the same color warmth and coldness. According to the engine area in each image, the color warmth and coldness of each grid of each image, and the corresponding total grid area, use the warmth and coldness weighted value analysis formula to determine the warmth and coldness weighted value of the engine;

[0010] Identify each image in the processed image set to obtain the S value and V value of the pixel points of each image. Analyze the average S value and average V value of each image according to the S value and V value of the pixel points of each image. Traverse the S value and V value of the pixel points of each image respectively. According to the average S value and average V value of the corresponding image, analyze the pixel points with color jumps in each image, and obtain the total number of color jump pixel points of each image. According to the total number of color jump pixel points of each image, use the color activity analysis formula to analyze the color activity of the engine;

[0011] Identify each image in the processed image set, and identify the parallel line groups that are parallel among the lines formed by the pipelines, lines, and components of the engine in each image. According to the number of parallel line groups in each image and the total number of line quantities of all parallel line groups in each image, use the regularity analysis method to analyze the regularity of the engine;

[0012] For each image in the processed image set, according to the horizontal symmetry axis of the engine, divide each symmetry axis dividing line of each image into the first area and the second area of each symmetry axis. Identify the number of characteristic blocks affecting the modeling balance sense, the area of each characteristic block, and the distance from the center of each characteristic block to the symmetry axis in the first area and the second area of each symmetry axis of each image. According to the number of characteristic blocks, the area of each characteristic block, and the distance from the center of each characteristic block to the symmetry axis in the first area and the second area of each symmetry axis of each image, use the balance degree analysis formula to analyze the balance degree of the engine;

[0013] According to the density, integrity, regularity, and balance degree of the engine, use the beauty degree analysis formula to analyze the beauty degree of the engine;

[0014] Output the design evaluation result according to the density, warmth and coldness weighted value, color activity, integrity, regularity, beauty degree, and balance degree.

[0015] In one of the embodiments, the expression of the density analysis formula is:

[0016]

[0017]

[0018] Among them, DM is the density of the engine, and DM j is the density of the j-th image of the engine, and a ij is the area of the i-th dense area of the j-th image of the engine, and a oj is the area of the engine in the j-th image of the engine. n is the number of dense areas in the j-th image of the engine, and p is the total number of images of the engine.

[0019] In one embodiment, the expression of the integrity analysis formula is:

[0020]

[0021] Among them, UM is the integrity of the engine, and S o is the area of the engine in the front image or the back image, and S g is the rectangular area of the minimum circumscribed rectangle of the engine in the front image or the back image.

[0022] In one embodiment, the expression of the cold and warm weighting value analysis formula is:

[0023]

[0024]

[0025] Among them, L is the cold and warm weighting value of the engine, and L j is the cold and warm weighting value of the j-th image of the engine, and l ij is the cold and warm degree of the color of the i-th grid of the j-th image of the engine, and S ij is the total area of the grids with the same cold and warm degree of color as the i-th grid in the j-th image of the engine, and S j总 is the area of the engine in the j-th image of the engine. m is the number of grids in the j-th image of the engine, and p is the total number of images of the engine.

[0026] In one embodiment, the expression of the color activity analysis formula is:

[0027]

[0028]

[0029] Among them, J is the color activity of the engine, and J j is the color activity of the j-th image of the engine, and v j is the total number of color-jumping pixel points in the j-th image of the engine.

[0030] In one of the embodiments, the expression of the regularity analysis method is:

[0031]

[0032]

[0033] where R j is the regularity of the j-th image of the engine, n gj is the number of parallel line groups in the j-th image of the engine, n lj is the total number of lines in all parallel line groups in the j-th image of the engine, R is the regularity of the engine, p is the total number of images of the engine, a ij is the area of the i-th dense area in the j-th image of the engine, a oj is the area of the engine in the j-th image of the engine, n is the number of dense areas in the j-th image of the engine, N + is a natural number.

[0034] In one of the embodiments, the expression of the balance analysis formula is:

[0035]

[0036]

[0037]

[0038]

[0039]

[0040]

[0041]

[0042]

[0043] where a ijU is the area of the i-th feature block in the first region of the horizontal symmetry axis of the j-th image of the engine, d ijUx is the distance from the center of the i-th feature block in the first region of the horizontal symmetry axis of the j-th image of the engine to the horizontal symmetry axis, F jUx is the visual strength value of the first region of the horizontal symmetry axis of the j-th image of the engine, w is the number of feature blocks in the first region of the horizontal symmetry axis of the j-th image of the engine, a ijDx is the area of the i-th feature block in the second region of the horizontal symmetry axis of the j-th image of the engine, dijDx is the distance from the center of the i-th feature block in the second region of the horizontal symmetry axis of the j-th image of the engine to the horizontal symmetry axis, F jDx is the visual intensity value of the second region of the horizontal symmetry axis of the j-th image of the engine, s is the number of feature blocks in the second region of the horizontal symmetry axis of the j-th image of the engine, B jx is the horizontal balance of the j-th image of the engine, a ijQy is the area of the i-th feature block in the first region of the vertical symmetry axis of the j-th image of the engine, d ijQy is the distance from the center of the i-th feature block in the first region of the vertical symmetry axis of the j-th image of the engine to the vertical symmetry axis, F jQy is the visual intensity value of the first region of the vertical symmetry axis of the j-th image of the engine, α is the number of feature blocks in the first region of the vertical symmetry axis of the j-th image of the engine, a ijAy is the area of the i-th feature block in the second region of the vertical symmetry axis of the j-th image of the engine, d ijAy is the distance from the center of the i-th feature block in the second region of the vertical symmetry axis of the j-th image of the engine to the vertical symmetry axis, F jAy is the visual intensity value of the second region of the vertical symmetry axis of the j-th image of the engine, β is the number of feature blocks in the second region of the vertical symmetry axis of the j-th image of the engine, B jy is the vertical balance of the j-th image of the engine, B j is the balance of the j-th image of the engine, B is the balance of the engine, p is the total number of images of the engine.

[0044] In one embodiment, the expression of the beauty degree analysis formula is:

[0045] M = DM×(0.1638R + 0.2973UM + 0.539B)

[0046] where, DM is the density of the engine, UM is the integrity of the engine, R is the regularity of the engine, B is the balance of the engine.

[0047] The above-mentioned engine design evaluation method based on machine vision collects images of different faces of the engine as the image set to be analyzed. The image set to be analyzed includes at least the front image and the back image of the engine. Preprocess the image set to be analyzed to obtain a processed image set. Perform image recognition and analysis on the processed image set, and use the density analysis formula to determine the density of the engine; perform recognition and analysis on the front image or the back image in the processed image set, and use the integrity analysis formula to determine the integrity of the engine; perform grid division and then recognition and analysis on each image in the processed image set, and use the warm and cold weighted value analysis formula to determine the warm and cold weighted value of the engine; perform recognition and analysis on each image in the processed image set, and use the color activity analysis formula to analyze the color activity of the engine; perform recognition and analysis on each image in the processed image set, and use the regularity analysis method to analyze the regularity of the engine; for each image in the processed image set, according to the horizontal symmetry axis of the engine, divide each image into a first region and a second region for each symmetry axis dividing line and perform recognition and analysis, and use the balance analysis formula to analyze the balance of the engine; according to the density, integrity, regularity and balance of the engine, use the beauty analysis formula to analyze the beauty of the engine; according to the density, warm and cold weighted value, color activity, integrity, regularity, beauty and balance, output the design evaluation result. Thus, the engine design evaluation can be more standardized, and further improve the accuracy of the engine design evaluation. Description of the Drawings

[0048] Figure 1 It is a schematic flow chart of an engine design evaluation method based on machine vision in an embodiment;

[0049] Figure 2 It is a schematic diagram of the dense area annotation of an image in an embodiment;

[0050] Figure 3 It is a schematic diagram of the image after annotating parallel lines in an embodiment. Detailed Description of the Embodiment

[0051] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0052] In one embodiment, as Figure 1 shown, a machine vision-based engine design evaluation method is provided. Taking the application of this method to a terminal as an example, the method includes the following steps:

[0053] Step 1, collect images of different surfaces of the engine as the image set to be analyzed, and the image set to be analyzed includes at least the front image and the back image of the engine.

[0054] Among them, images of different surfaces of the engine can be collected as needed.

[0055] Step 2, preprocess the image set to be analyzed to obtain a processed image set.

[0056] Among them, the preprocessing can be to perform processing such as background subtraction on the images in the image set to be analyzed. For example: perform grayscale processing on the images and normalize the background pixel values according to a threshold.

[0057] Step 3, perform image recognition on the processed image set to obtain the engine area of each image in the processed image set, and the areas of the dense regions of the pipelines, lines, and components in each image. According to the engine area of each image and the areas of each dense region, use the density analysis formula to determine the density of the engine.

[0058] Among them, the method of identifying the dense regions of the pipelines, lines, and components can be to use a neural network model to recognize the images and identify the dense regions of the pipelines, lines, and components, and then obtain the areas of the dense regions. It can also be to pre-label the dense regions of the pipelines, lines, and components in the images of the processed image set in advance, and then identify the areas of the labeled dense regions.

[0059] In one embodiment, the dense regions of the pipelines, lines, and components in the images of the processed image set are pre-labeled with color blocks, and the colors shown in the HSV and color temperature correspondence table in Table 1 can be used for labeling, with the range: (H value is 0 to 180, S value is 0 to 255, V value is 0 to 255). In one example, as Figure 2 shown in the schematic diagram of the dense region labeling of the image, when processing the image, the selected color is red, and the H value, S value, and V value are respectively: 10±10, 245±10, and 245±10.

[0060] Table 1 HSV and color temperature correspondence table

[0061] H value (hue) S (saturation) V (brightness) Color <![CDATA[Color warmth or coolness (L i )]]> 0 to 10 and 156 to 180 43～255 46～255 Red 0 11～25 43～255 46～255 Orange 1 / 6 26～34 43～255 46～255 Yellow 2 / 6 35～77 43～255 46～255 Green 4 / 6 78～99 43～255 46～255 Cyan 1 100～124 43～255 46～255 Blue 5 / 6 125～155 43～255 46～255 Purple 3 / 6 0～180 0～255 0～46 Black 3 / 6 0～180 0～43 46～220 Gray 3 / 6 0～180 0～30 221～255 White 3 / 6

[0062] In one embodiment, the expression of the density analysis formula is:

[0063]

[0064]

[0065] Among them, DM is the density of the engine, DMj is the density of the j-th image of the engine, a ij is the area of the i-th dense region of the j-th image of the engine, a oj is the area of the engine in the j-th image of the engine, n is the number of dense regions in the j-th image of the engine, and p is the total number of images of the engine.

[0066] Step 4: Identify the front image or the back image in the processed image set, obtain the rectangular area of the minimum circumscribed rectangle of the engine in the front image or the back image, and the area of the engine in the front image or the back image. According to the rectangular area and the engine area in the front image or the back image, use the integrity analysis formula to determine the integrity of the engine.

[0067] Among them, the rectangular area of the minimum circumscribed rectangle of the engine can be obtained by cropping the image in the processed image set with the minimum circumscribed rectangle of the engine, and identifying the area of the cropped image as the rectangular area.

[0068] In one embodiment, the expression of the integrity analysis formula is:

[0069]

[0070] Among them, UM is the integrity of the engine, S o is the area of the engine in the front image or the back image, S g is the rectangular area of the minimum circumscribed rectangle of the engine in the front image or the back image.

[0071] Step 5: After dividing each image in the processed image set into grids, identify the HSV average value of each grid of each image and the area of the engine in each image. According to the HSV average value of each grid of each image, determine the color warmth and coldness of each grid of each image and the total area of the grids with the same color warmth and coldness. According to the area of the engine in each image, the color warmth and coldness of each grid of each image, and the corresponding total area of the grids, use the warm and cold weighted value analysis formula to determine the warm and cold weighted value of the engine.

[0072] Among them, when dividing the image into grids, the number of grids can be determined as needed. In one example, it can be divided into 30 grids horizontally and 10 grids vertically.

[0073] Among them, after dividing each image in the processed image set into grids, the image can also be processed to remove color jumps (color jumps refer to relatively bright colors, such as the whole is black and one line is red), and then the HSV average value of each grid of the image can be identified and analyzed.

[0074] Among them, the HSV average value of the grid includes the average H value, average S value, and average V value of the grid. The average H value of the grid can be obtained by averaging the H values of each pixel point in the grid. The average S value of the grid can be obtained by averaging the S values of each pixel point in the grid. The average V value of the grid can be obtained by averaging the V values of each pixel point in the grid.

[0075] In one embodiment, the expression of the cold and warm weighting value analysis formula is:

[0076]

[0077]

[0078] Among them, L is the cold and warm weighting value of the engine, L j is the cold and warm weighting value of the j-th image of the engine, l ij is the color cold and warmth degree of the i-th grid in the j-th image of the engine, S ij is the total area of the grids with the same color cold and warmth degree as the i-th grid in the j-th image of the engine, S j总 is the area of the engine in the j-th image of the engine, m is the number of grids in the j-th image of the engine, and p is the total number of images of the engine.

[0079] Step 6: Identify each image in the processed image set to obtain the S value and V value of each pixel point of each image. Analyze the average S value and average V value of each image based on the S value and V value of each pixel point of each image. Traverse the S value and V value of each pixel point of each image respectively. Based on the average S value and average V value of the corresponding image, analyze the pixel points with color jumps in each image to obtain the total number of color jump pixel points of each image. Based on the total number of color jump pixel points of each image, analyze the color activity degree of the engine using the color activity degree analysis formula.

[0080] In one embodiment, the judgment condition for analyzing the pixel points with color jumps in each image is: both the S value and V value of the pixel point are within the range shown in Table 1, and if the following Condition 1 and Condition 2 are satisfied simultaneously, it is determined as a pixel point with color jumps.

[0081] Condition 1: The difference between the S value S z of the z-th pixel point of this image and the average S value S 平 of this image is greater than the preset S value threshold.

[0082] Condition 2: The difference between the V z of the z-th pixel point of this image and the average V value V 平 of this image is greater than the preset V value threshold.

[0083] Among them, the preset S value threshold and the preset V value threshold are appropriately changed according to the result of screening for color skipping.

[0084] In one embodiment, the expression of the color activity analysis formula is:

[0085]

[0086]

[0087] Among them, J is the color activity of the engine, and J j is the color activity of the j-th image of the engine, and v j is the total number of color-skipping pixel points in the j-th image of the engine.

[0088] Step 7: Identify each image in the processed image set, identify the parallel line groups formed by the pipelines, lines, and components of the engine in the lines formed by them, and analyze the regularity of the engine by using the regularity analysis method according to the number of parallel line groups in each image and the total number of lines in all parallel line groups in each image.

[0089] Among them, for the parallel line groups formed by the pipelines, lines, and components in the lines, the method can be to use a neural network model to identify the parallel lines in the lines formed by the pipelines, lines, and components, and then identify the parallel line groups formed by the pipelines, lines, and components of the engine in the lines formed by them in each image, as well as the total number of lines in all parallel line groups in the image. It can also be to pre-label the parallel lines of the pipelines, lines, and components in the images of the processed image set, and then identify the parallel line groups formed by the pipelines, lines, and components of the engine in the lines formed by them in each image, as well as the total number of lines in all parallel line groups in the image.

[0090] In one embodiment, the parallel lines of the pipelines, lines, and components in the images of the processed image set can be pre-labeled with color blocks, and the colors shown in the color suggestion table in Table 2 can be used for labeling, such as Figure 3 the image after labeling the parallel lines as shown.

[0091] Table 2 Color Suggestion Table

[0092] Red: H, S, V (5 ± 5, 250 ± 5, 250 ± 5) Orange: H, S, V (20 ± 5, 250 ± 5, 250 ± 5) Yellow: H, S, V (30 ± 5, 250 ± 5, 250 ± 5) Green: H, S, V (70 ± 5, 250 ± 5, 250 ± 5) Cyan: H, S, V (90 ± 5, 250 ± 5, 250 ± 5) Blue: H, S, V (120 ± 5, 250 ± 5, 250 ± 5) Purple: H, S, V (150 ± 5, 250 ± 5, 250 ± 5)

[0093] In one embodiment, the expression of the regularity analysis method is:

[0094]

[0095]

[0096] Among them, R j is the regularity of the j-th image of the engine, n gj is the number of parallel line groups in the j-th image of the engine, n lj is the total number of lines in all parallel line groups in the j-th image of the engine, R is the regularity of the engine, p is the total number of images of the engine, a ij is the area of the i-th dense area in the j-th image of the engine, a oj is the engine area of the j-th image of the engine, n is the number of dense areas in the j-th image of the engine, N + is a natural number.

[0097] It should be understood that when there are no parallel lines in the image, the value of n lj can be taken as 1.

[0098] Step 8, for each image in the processed image set, according to the horizontal symmetry axis of the engine, divide each image by each symmetry axis dividing line into a first area and a second area of each symmetry axis, identify the number of characteristic blocks affecting the styling balance, the area of each characteristic block, and the distance from the center of each characteristic block to the symmetry axis in the first area and the second area of each symmetry axis of each image, and analyze the balance of the engine by using the balance analysis formula according to the number of characteristic blocks, the area of each characteristic block, and the distance from the center of each characteristic block to the symmetry axis in the first area and the second area of each symmetry axis of each image.

[0099] Among them, the characteristic block affecting the styling balance can be an asymmetric area.

[0100] Among them, before dividing each image by each symmetry axis dividing line into a first area and a second area of each symmetry axis, the image can be recognized by using a neural network model to recognize the horizontal symmetry axis and the vertical symmetry axis of each image, and then divide each image by each symmetry axis dividing line into a first area and a second area of each symmetry axis; or the horizontal symmetry axis and the vertical symmetry axis of each image can be marked in advance, and then divide each image by each symmetry axis dividing line into a first area and a second area of each symmetry axis.

[0101] In one embodiment, the number and type of symmetry axes can also be determined according to the type of the engine. For example, the engines of fighter planes can analyze the balance of the engine by using two symmetry axes, namely the horizontal symmetry axis and the vertical symmetry axis; commercial engines can analyze the balance of the engine only by using the horizontal symmetry axis.

[0102] In one embodiment, the expression of the balance analysis formula is:

[0103]

[0104]

[0105]

[0106]

[0107]

[0108]

[0109]

[0110]

[0111] where a ijUx is the area of the i-th feature block in the first region of the horizontal symmetry axis of the j-th image of the engine, and d ijUx is the distance from the center of the i-th feature block in the first region of the horizontal symmetry axis of the j-th image of the engine to the horizontal symmetry axis, and F jUx is the visual strength value of the first region of the horizontal symmetry axis of the j-th image of the engine, w is the number of feature blocks in the first region of the horizontal symmetry axis of the j-th image of the engine, and a ijDx is the area of the i-th feature block in the second region of the horizontal symmetry axis of the j-th image of the engine, and d ijDx is the distance from the center of the i-th feature block in the second region of the horizontal symmetry axis of the j-th image of the engine to the horizontal symmetry axis, and F jQx is the visual strength value of the second region of the horizontal symmetry axis of the j-th image of the engine, s is the number of feature blocks in the second region of the horizontal symmetry axis of the j-th image of the engine, and B jx is the horizontal balance of the j-th image of the engine, and a ijQy is the area of the i-th feature block in the first region of the vertical symmetry axis of the j-th image of the engine, and d ijQy is the distance from the center of the i-th feature block in the first region of the vertical symmetry axis of the j-th image of the engine to the vertical symmetry axis, and Fj jQy is the visual strength value of the first region of the vertical symmetry axis of the j-th image of the engine, α is the number of feature blocks in the first region of the vertical symmetry axis of the j-th image of the engine, and a ijAy is the area of the i-th feature block in the second region of the vertical symmetry axis of the j-th image of the engine, and d ijAy is the distance from the center of the i-th feature block in the second region of the vertical symmetry axis of the j-th image of the engine to the vertical symmetry axis, and F jAyis the visual intensity value of the second region of the vertical symmetry axis of the j-th image of the engine, β is the number of characteristic blocks in the second region of the vertical symmetry axis of the j-th image of the engine, B jy is the vertical balance of the j-th image of the engine, B j is the balance of the j-th image of the engine, B is the balance of the engine, and p is the total number of images of the engine.

[0112] It is understood that in the case of analyzing the balance of the engine using the horizontal symmetry axis, there is no need to analyze the vertical balance of the image.

[0113] Step 9: Analyze the beauty degree of the engine using the beauty degree analysis formula based on the density, integrity, regularity, and balance of the engine.

[0114] The expression of the beauty degree analysis formula is:

[0115] M = DM × (0.1638R + 0.2973UM + 0.539B)

[0116] where DM is the density of the engine, UM is the integrity of the engine, R is the regularity of the engine, and B is the balance of the engine.

[0117] Step 10: Output the design evaluation result based on the density, cold and warm weighting value, color activity, integrity, regularity, beauty degree, and balance.

[0118] The above-mentioned engine design evaluation method based on machine vision collects images of different surfaces of the engine as the image set to be analyzed. The image set to be analyzed includes at least the front image and the back image of the engine. The image set to be analyzed is preprocessed to obtain a processed image set. Image recognition is performed on the processed image set to obtain the engine area of each image in the processed image set, and the areas of the dense regions of pipelines, circuits, and components in each image. According to the engine area of each image and the area of each dense region, the density of the engine is determined using the density analysis formula; Image recognition is performed on the front image or the back image in the processed image set to obtain the rectangular area of the minimum circumscribed rectangle of the engine in the front image or the back image, and the engine area in the front image or the back image. According to the rectangular area and the engine area in the front image or the back image, the integrity of the engine is determined using the integrity analysis formula; After each image in the processed image set is grid-divided, the HSV average value of each grid of each image and the engine area in each image are recognized. According to the HSV average value of each grid of each image, the color warmth and coldness of each grid of each image and the total area of the grids with the same color warmth and coldness are determined. According to the engine area in each image, the color warmth and coldness of each grid of each image, and the corresponding total grid area, the warmth and coldness weighted value of the engine is determined using the warmth and coldness weighted value analysis formula; Image recognition is performed on each image in the processed image set to obtain the S value and the V value of the pixel points of each image. The average S value and the average V value of each image are analyzed based on the S value and the V value of the pixel points of each image. The pixel points with color jumps in each image are analyzed by traversing the S value and the V value of the pixel points of the corresponding image. The total number of pixel points with color jumps in each image is obtained. According to the total number of pixel points with color jumps in each image, the color activity of the engine is analyzed using the color activity analysis formula; Image recognition is performed on each image in the processed image set to identify the parallel line groups formed by the lines of the pipelines, circuits, and components of the engine in each image. According to the number of parallel line groups in each image and the total number of line quantities of all parallel line groups in each image, the regularity of the engine is analyzed using the regularity analysis method;For each image in the processed image set, divide each image along the horizontal symmetry axis of the engine, and divide each symmetry axis dividing line of each image into a first region and a second region of each symmetry axis. Identify the number of characteristic blocks affecting the styling balance, the area of each characteristic block, and the distance from the center of each characteristic block to the symmetry axis in the first region and the second region of each symmetry axis in each image. According to the number of characteristic blocks, the area of each characteristic block, and the distance from the center of each characteristic block to the symmetry axis in the first region and the second region of each symmetry axis in each image, use the balance degree analysis formula to analyze the balance degree of the engine; according to the density, integrity, regularity, and balance degree of the engine, use the beauty degree analysis formula to analyze the beauty degree of the engine; according to the density, cold and warm weighting value, color activity, integrity, regularity, beauty degree, and balance degree, output the design evaluation result. Thus, the engine design evaluation can be more standardized, thereby improving the accuracy of the engine design evaluation. It can also greatly improve the calculation speed and effectively improve the work efficiency. A detection system can also be added to judge whether the input is compliant, whether the result is correct, etc.;

[0119] It should be understood that although the step numbers of steps 1 - 10 of the above-mentioned engine design evaluation method based on machine vision are numbered. Unless there is a clear description in this article, the execution of these steps does not have a strict order limit, and these steps can be executed in other orders. Moreover, at least some of the steps 1 - 10 of the engine design evaluation method based on machine vision can include multiple sub-steps or multiple stages. These sub-steps or stages do not necessarily need to be executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed alternately or alternately with at least a part of other steps or sub-steps or stages of other steps.

[0120] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.

[0121] The above-mentioned embodiments only represent several implementation manners of the present application. The description is relatively specific and detailed, but it should not be understood as a limitation to the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several deformations and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. An engine design evaluation method based on machine vision, characterized in that, The method includes: Collecting images of different surfaces of the engine as an image set to be analyzed, where the image set to be analyzed includes at least a front image and a back image of the engine; Performing preprocessing on the image set to be analyzed to obtain a processed image set; Performing image recognition on the processed image set to obtain the engine area of each image in the processed image set, and the areas of dense regions of pipelines, circuits, and components in each image. According to the engine area and the area of each dense region of each image, the density of the engine is determined using a density analysis formula; Performing recognition on the front image or the back image in the processed image set to obtain the rectangular area of the minimum circumscribed rectangle of the engine in the front image or the back image, and the engine area in the front image or the back image. According to the rectangular area and the engine area in the front image or the back image, the integrity of the engine is determined using an integrity analysis formula; After performing grid division on each image in the processed image set, recognizing the HSV average values of each grid of each image and the engine area in each image. According to the HSV average values of each grid of each image, determining the color warmth and coldness of each grid of each image and the total area of the grids with the same color warmth and coldness. According to the engine area in each image, the color warmth and coldness of each grid of each image, and the corresponding total grid area, the warmth and coldness weighted value of the engine is determined using a warmth and coldness weighted value analysis formula; Performing recognition on each image in the processed image set to obtain the S value and V value of the pixel points of each image. Analyzing the average S value and average V value of each image according to the S value and V value of the pixel points of each image. Traversing the S value and V value of the pixel points of each image respectively. According to the average S value and average V value of the corresponding image, analyzing the pixel points with color jumps in each image to obtain the total number of color jump pixel points of each image. According to the total number of color jump pixel points of each image, analyzing the color activity of the engine using a color activity analysis formula; Performing recognition on each image in the processed image set to recognize the parallel line groups of parallel lines formed by the pipelines, circuits, and components of the engine in each image. According to the number of parallel line groups in each image and the total number of line quantities of all parallel line groups in each image, analyzing the regularity of the engine using a regularity analysis method; For each image in the processed image set, according to the horizontal symmetry axis of the engine, dividing each image by each symmetry axis dividing line into a first region and a second region of each symmetry axis. Recognizing the number of characteristic blocks affecting the modeling balance sense, the area of each characteristic block, and the distance from the center of each characteristic block to the symmetry axis in the first region and the second region of each symmetry axis of each image. According to the number of characteristic blocks, the area of each characteristic block, and the distance from the center of each characteristic block to the symmetry axis in the first region and the second region of each symmetry axis of each image, analyzing the balance of the engine using a balance analysis formula; According to the density, integrity, regularity, and balance of the engine, the beauty degree of the engine is analyzed using the beauty degree analysis formula; Based on the density, cold and warm weighting value, color activity, integrity, regularity, beauty degree, and balance, the design evaluation result is output.

2. The method according to claim 1, wherein The expression of the density analysis formula is: Among them, DM is the density of the engine, and DM j is the density of the j-th image of the engine, and a ij is the area of the i-th dense region of the j-th image of the engine, and a oj is the area of the engine in the j-th image of the engine, n is the number of dense regions in the j-th image of the engine, and p is the total number of images of the engine.

3. The method according to claim 1, wherein The expression of the integrity analysis formula is: Among them, UM is the integrity of the engine, and S o is the area of the engine in the front image or the back image, and S g is the rectangular area of the minimum circumscribed rectangle of the engine in the front image or the back image.

4. The method according to claim 1, wherein The expression of the cold and warm weighting value analysis formula is: Among them, L is the cold and warm weighted value of the engine, L j is the cold and warm weighted value of the j-th image of the engine, l ij is the color cold and warmth degree of the i-th grid of the j-th image of the engine, S ij is the total area of the grids with the same color cold and warmth degree as the i-th grid in the j-th image of the engine, S j总 is the area of the engine in the j-th image of the engine, m is the number of grids in the j-th image of the engine, and p is the total number of images of the engine.

5. The method according to claim 1, wherein The expression of the color activity analysis formula is: Among them, J is the color activity of the engine, and J j is the color activity of the j-th image of the engine, and v j is the total number of pixel points with color jumps in the j-th image of the engine, and p is the total number of images of the engine.

6. The method according to claim 1, characterized in that, The expression of the regularity analysis method is: Among them, R j is the regularity of the j-th image of the engine, n gj is the number of parallel line groups in the j-th image of the engine, n lj is the total number of lines in all parallel line groups of the j-th image of the engine, R is the regularity of the engine, p is the total number of images of the engine, a ij is the area of the i-th dense region in the j-th image of the engine, a oj is the engine area of the j-th image of the engine, n is the number of dense regions in the j-th image of the engine, N + is a natural number.

7. The method according to claim 1, characterized in that The expression of the balance analysis formula is: where a ijUx is the area of the i-th feature block in the first region of the horizontal symmetry axis of the j-th image of the engine, d ijUx is the distance from the center of the i-th feature block in the first region of the horizontal symmetry axis of the j-th image of the engine to the horizontal symmetry axis, F jUx is the visual strength value of the first region of the horizontal symmetry axis of the j-th image of the engine, w is the number of feature blocks in the first region of the horizontal symmetry axis of the j-th image of the engine, a ijDx is the area of the i-th feature block in the second region of the horizontal symmetry axis of the j-th image of the engine, d ijDx is the distance from the center of the i-th feature block in the second region of the horizontal symmetry axis of the j-th image of the engine to the horizontal symmetry axis, F jDx is the visual strength value of the second region of the horizontal symmetry axis of the j-th image of the engine, s is the number of feature blocks in the second region of the horizontal symmetry axis of the j-th image of the engine, B jx is the horizontal balance degree of the j-th image of the engine, a ijQy is the area of the i-th feature block in the first region of the vertical symmetry axis of the j-th image of the engine, d ijQy is the distance from the center of the i-th feature block in the first region of the vertical symmetry axis of the j-th image of the engine to the vertical symmetry axis, F jQy is the visual strength value of the first region of the vertical symmetry axis of the j-th image of the engine, α is the number of feature blocks in the first region of the vertical symmetry axis of the j-th image of the engine, a ijAy is the area of the i-th feature block in the second region of the vertical symmetry axis of the j-th image of the engine, d ijAy is the distance from the center of the i-th feature block in the second region of the vertical symmetry axis of the j-th image of the engine to the vertical symmetry axis, F jAy is the visual strength value of the second region of the vertical symmetry axis of the j-th image of the engine, β is the number of feature blocks in the second region of the vertical symmetry axis of the j-th image of the engine, B jy is the vertical balance degree of the j-th image of the engine, B j is the balance degree of the j-th image of the engine, B is the balance degree of the engine, and p is the total number of images of the engine.

8. The method according to claim 1, characterized in that, The expression of the beauty degree analysis formula is: M = DM × (0.1638R + 0.2973UM + 0.539B) Where DM is the density of the engine, UM is the integrity of the engine, R is the regularity of the engine, and B is the balance of the engine.

Images