WO2018032631A1 - 一种教育玩具套件及其电路元件和电线的识别方法 - Google Patents

一种教育玩具套件及其电路元件和电线的识别方法 Download PDF

Info

Publication number
WO2018032631A1
WO2018032631A1 PCT/CN2016/105741 CN2016105741W WO2018032631A1 WO 2018032631 A1 WO2018032631 A1 WO 2018032631A1 CN 2016105741 W CN2016105741 W CN 2016105741W WO 2018032631 A1 WO2018032631 A1 WO 2018032631A1
Authority
WO
WIPO (PCT)
Prior art keywords
color
image
circuit component
wire
circuit
Prior art date
Application number
PCT/CN2016/105741
Other languages
English (en)
French (fr)
Inventor
范旭
程潇
杜鹏
暴满粟
孙贤军
Original Assignee
上海葡萄纬度科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 上海葡萄纬度科技有限公司 filed Critical 上海葡萄纬度科技有限公司
Publication of WO2018032631A1 publication Critical patent/WO2018032631A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/10Segmentation; Edge detection
    • G06T7/12Edge-based segmentation
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/0002Inspection of images, e.g. flaw detection

Definitions

  • the invention relates to the technical field of computer vision detection and processing, in particular to an educational toy kit and a circuit element and a wire identification method thereof.
  • an educational toy kit has been successfully developed in the field of computer vision and image processing technology, including: a bracket, a helmet detector and a bottom plate, and a game program is installed in the tablet, and is collected on a flat surface by a tablet computer. The image of the bottom plate.
  • the present invention provides an educational toy kit and a circuit component and a wire identification method thereof.
  • the technical solution is as follows:
  • An educational toy kit comprising: a bottom plate, circuit components and a wire bottom plate placed on a flat surface, and circuit components and wires placed on the bottom plate.
  • the bottom plate is a rectangle having rounded corners, and a calibration angle is provided at four corners of the rectangle.
  • the calibration angle is a red circular arc.
  • a method for identifying circuit components and wires in an educational toy kit includes the following steps:
  • Step one install the game program on the tablet, and then place the bottom plate on the plane to ensure that the side of the calibration angle is facing up.
  • Step 2 complete the connection of the circuit components and the wires on the bottom plate, collect the color images in real time through the rear camera of the tablet computer, and move the tablet computer to ensure that the color images collected by the rear camera have at least three calibration angles;
  • Step 3 extracting a valid identification area from the color image of step 2;
  • Step 4 detecting circuit components located in the effective recognition area of the color image
  • Step 5 detecting a wire located in an effective recognition area of the color image
  • Step 6 Determine whether the connection between the circuit components and the wires is accurate.
  • the color value, G xy represents the color value of the image pixel in the green channel, and B xy represents the color value of the image pixel in the blue channel.
  • the specific steps of extracting the effective recognition region from the color image in step 3 are:
  • step B) scanning the four calibration angle binary images obtained in step A) to obtain the corresponding edge contour map, and filtering out the unreasonable contour according to the prior knowledge of the eccentricity and size of the edge contour;
  • step C) Calculate the circumscribed rectangles of the four calibration angles according to the remaining edge contours obtained in step B). In the identification process, when there are at least three angle markers with matching calibration angles, the circumscribed rectangle is calculated. Effective identification area.
  • the step of detecting the circuit components located in the color image recognition region in step 4 is:
  • the specific steps of extracting the inner contour of each circuit component housing in step 1 are:
  • the extracted region of interest image is converted from the RGB color space to the HSV color space focusing on the color representation.
  • the specific conversion formula is:
  • H is the tone value
  • S is the saturation value
  • V is the brightness value
  • max ⁇ C(R'), C(G'), C(B') ⁇ means that one pixel is in red and green in the original image.
  • the maximum pixel value of the three channels of blue, min ⁇ C(R'), C(G'), C(B') ⁇ indicates that the pixel of one pixel in the original image is the smallest in the three channels of red, green and blue.
  • Value, and the value range of H is between 0-360;
  • B(x, y) represents the binary pixel value of the image pixel point (x, y)
  • H(x, y) S(x, y), V(x, y) respectively represent the image pixel point (x, y) the hue value, saturation value, and brightness value in the HSV color space
  • B_H(x, y), B_S(x, y), B_V(x, y) respectively indicate whether the image pixel points (x, y) are respectively In the specified H, S, and V regions, if yes, the value is 1, otherwise, the value is 0
  • H min and H max respectively indicate the a priori color of the color of a component shell in the HSV color space.
  • the minimum and maximum values; S min and S max respectively represent the a priori minimum and maximum values of the saturation of the color of a component shell in the HSV color space; V min and V max respectively indicate the color of a component shell in the HSV. A priori minimum and maximum values of brightness in the color space.
  • the binarized image can be regarded as a grayscale image with only two values.
  • the edge of the image refers to the part of the grayscale image where the grayscale changes are more severe.
  • the degree of change of the grayscale value is quantified by the gradient change between adjacent pixels.
  • the gradient is a two-dimensional equivalent of the first-order two-dimensional derivative.
  • G x represents the difference of adjacent pixels in the x direction
  • G y represents the difference of adjacent pixels in the y direction
  • f[i, j+1] represents the pixel value of the image in the i th row and j+1th column.
  • f[i,j] represents the pixel value of the image in the i-th row and the j-th column
  • f[i+1,j] represents the pixel value of the image in the i-th row and the j-th column;
  • G(x, y) represents the gradient value at the (x, y) point of the image
  • the gradient magnitude of the edge point is calculated, and the gradient magnitude set of all the edge points is the extracted edge contour;
  • the method of calculating the non-stationary electric fan and the remaining circuit component ears is based on the prior knowledge of the shape, size and eccentricity of the circuit component's ear contour, filtering out the unreasonable contour, thereby obtaining the two ears of the circuit component. s position;
  • step b) needs to be color-divided to segment the blue binary image and the green binary image;
  • the blue binary image is scanned, and the prior knowledge of the contour shape, the eccentricity, and the size of the blade is used to determine whether there is a blade, and if so, the circuit component is determined to be a fan; otherwise, Skip directly to step 2;
  • the calculation method for the ear of the static fan is as follows: clustering the green binary image in step 1 to cluster the green dots closer to each other, through the same class
  • the number of point sets, the size of the minimum outsourcing rectangle of the point set, the eccentricity of the point set, and the prior knowledge of the point set position are used to filter the set of points that are clustered into one class. If only two types of point sets remain after filtering, the current fan blade is considered
  • the position has a large probability to cover the ears of the electrical components, and the center point position of the ear of the circuit component is calculated by the center point position of the two types of point sets and the contour information of the components.
  • the specific steps of calculating the position and the deflection angle of each circuit component in step 2 are:
  • the rotation angle of the circuit component is calculated by ensuring that the ear of the circuit component is in the horizontal direction.
  • the specific steps of identifying the circuit component category in step 3 are:
  • the private printing color in the upper middle, middle middle and lower middle red is 1, yellow is 2, blue is 3, green is 4, no is 0, and the current color is ignored.
  • 0, 11 circuit components can be encoded, according to the circuit component code, the type of the current detection circuit component can be uniquely determined;
  • circuit component codes are as follows:
  • the color pixel threshold is 30% of the area of each of the nine squares.
  • the specific steps of detecting the wires located in the effective recognition area of the color image in step 5 are:
  • Step 1) dividing the wire and extracting the bone by the color difference
  • Step 2) through the extracted bones, calculate the end points and bifurcation points of the wires.
  • the specific steps of dividing the wire and extracting the bone by the color difference in the step 1) are:
  • the image is binarized according to the a priori threshold value of the color of the wire in the HSV space to obtain a wire binarized image
  • the wire binarized image is scanned, and the contour of the wire profile is filtered by a priori knowledge of the shape and size of the wire profile, and the remaining contour is filled to obtain a new wire-only binarization map. ;
  • the above-described wire-only binarization map is subjected to bone refinement to obtain the skeleton of the wire.
  • the specific steps of calculating the end points and the branch points of the wires by the extracted bones in step 2) are: obtained by step 1) Wire skeleton, as well as a priori knowledge of the endpoints and bifurcation characteristics of the wire, find the endpoints and bifurcation points in the wire skeleton, and if there are bifurcation points, separate the wire skeleton from the bifurcation point and divide it into multiple segments.
  • the original wire bone is a line segment that provides all the line segments and the type of segment end in the bone to the upper layer software.
  • the specific step of determining whether the connection of the circuit component and the wire is accurate in step 6 is: the circuit element identified by the upper layer software in step four The device is connected to the wire identified in step 5, and then compared with the circuit diagram of the upper layer software for storage. If it is consistent with the circuit diagram, the circuit connection is considered to be accurate.
  • the invention intelligently combines the application of computer vision pattern recognition technology with HSV color space, binarization processing, and image cutting technology, can determine the type of circuit components and wires, and can determine whether the circuit connection is accurate. It has fast computing speed and accurate positioning. It integrates hardware and software technology well.
  • the game interaction design is ingenious. It is simple and beautiful, and the judgment is faster. It enhances the child's imagination and increases the fun of the game, so that children can learn the basic circuit knowledge. To cultivate children's interests.
  • the detection algorithm of the invention is more scientific and mature, and combines image color conversion, image convolution, image cutting, bone refinement and the like to quickly determine the type of circuit components and wires.
  • the calculation speed of the invention is fast; each positioning detection takes about 200ms, which provides a smooth experience for the player.
  • the performance of the invention is stable. In the case of different illumination and different tablet computers installed in the educational toy kit, the collection and test of 3,000 pictures are performed, and the false recognition rate and the missed detection rate are below 0.2%.
  • Figure 1 is a schematic view showing the structure of an educational toy kit of the present invention.
  • FIG. 2 is a flow chart of a method of identifying circuit components and wires in an educational toy kit of the present invention.
  • a method for identifying circuit components and wires in an educational toy kit includes the following steps:
  • Step 1 Install the game program on the tablet, and then place the bottom plate on the plane to ensure that the side of the calibration angle is facing up;
  • Step 2 complete the connection between the circuit components and the wires on the bottom plate, collect the color images in real time through the rear camera of the tablet computer, and move the tablet computer to ensure that the color images collected by the rear camera have at least three calibration angles, the specific steps are :
  • Step 3 extracting a valid identification area from the color image of step 2, the specific steps are:
  • step B) scanning the four calibration angle binary images obtained in step A) to obtain the corresponding edge contour map, and filtering out the unreasonable contour according to the prior knowledge of the eccentricity and size of the edge contour;
  • Step 4 detecting circuit components located in the effective recognition area of the color image, the specific steps are:
  • the extracted region of interest image is converted from the RGB color space to the HSV color space focusing on the color representation.
  • the specific conversion formula is:
  • H is the tone value
  • S is the saturation value
  • V is the brightness value
  • max ⁇ C(R'), C(G'), C(B') ⁇ means that one pixel is in red and green in the original image.
  • the maximum pixel value of the three channels of blue, min ⁇ C(R'), C(G'), C(B') ⁇ indicates that the pixel of one pixel in the original image is the smallest in the three channels of red, green and blue.
  • Value, and the value range of H is between 0-360;
  • the color image is binarized according to the a priori threshold in the HSV space of the color of the circuit component's outer casing.
  • the specific formula is as follows:
  • B(x, y) represents the binary pixel value of the image pixel point (x, y)
  • H(x, y) S(x, y), V(x, y) respectively represent the image pixel point (x, y) the hue value, saturation value, and brightness value in the HSV color space
  • B_H(x, y), B_S(x, y), B_V(x, y) respectively indicate whether the image pixel points (x, y) are respectively In the specified H, S, and V regions, if yes, the value is 1, otherwise, the value is 0
  • H min and H max respectively indicate the a priori color of the color of a component shell in the HSV color space.
  • S min and S max respectively represent the a priori minimum and maximum values of the saturation of the color of a component shell in the HSV color space
  • V min and V max respectively indicate the color of a component shell in the HSV.
  • the binarized image can be regarded as only two grayscale images.
  • the edge of the image refers to the part of the grayscale image where the grayscale changes are more intense.
  • the degree of change of the grayscale value is quantified by the gradient change between adjacent pixels.
  • the gradient is a two-dimensional equivalent of the first-order two-dimensional derivative.
  • G x represents the difference of adjacent pixels in the x direction
  • G y represents the difference of adjacent pixels in the y direction
  • f[i, j+1] represents the pixel value of the image in the i th row and j+1th column.
  • f[i,j] represents the pixel value of the image in the i-th row and the j-th column
  • f[i+1,j] represents the pixel value of the image in the i-th row and the j-th column
  • G(x, y) represents the gradient value at the (x, y) point of the image
  • the gradient magnitude of the edge point is calculated, and the gradient magnitude set of all the edge points is the extracted edge contour;
  • the method of calculating the non-stationary electric fan and the remaining circuit component ears is based on the prior knowledge of the shape, size and eccentricity of the circuit component's ear contour, filtering out the unreasonable contour, thereby obtaining the two ears of the circuit component. s position;
  • step b) needs to be color-divided to segment the blue binary image and the green binary image;
  • the blue binary image is scanned, and the prior knowledge of the contour shape, the eccentricity, and the size of the blade is used to determine whether there is a blade, and if so, the circuit component is determined to be a fan; otherwise, Skip directly to step 2;
  • the calculation method for the ear of the static fan is as follows: clustering the green binary image in step 1 to cluster the green dots closer to each other, through the same class
  • the number of point sets, the size of the minimum outsourcing rectangle of the point set, the eccentricity of the point set, and the prior knowledge of the point set position are used to filter the set of points that are clustered into one class. If only two types of point sets remain after filtering, the current fan blade is considered
  • the position has a large probability to cover the ear of the electrical component, and the center point position of the ear of the circuit component is calculated by the center point position of the two types of point sets and the contour information of the component;
  • the rotation angle of the circuit component is calculated by ensuring that the ear of the circuit component is in the horizontal direction. According to the calculated ear position and center position of the circuit component, to ensure that the ear of the circuit component is in the horizontal direction, the rotation angle of the circuit component is calculated;
  • the private printing color in the upper middle, middle middle and lower middle red is 1, yellow is 2, blue is 3, green is 4, no is 0, and the current color is ignored.
  • 0, 11 circuit components can be encoded, according to the circuit component code, the type of the current detection circuit component can be uniquely determined;
  • circuit component codes are as follows:
  • Step 5 detecting the wires located in the effective recognition area of the color image, the specific steps are:
  • Step 1) by dividing the wire and extracting the bone by color difference, the specific steps are as follows:
  • the image is binarized according to the a priori threshold value of the color of the wire in the HSV space to obtain a wire binarized image
  • the wire binarized image is scanned, and the contour of the wire profile is filtered by a priori knowledge of the shape and size of the wire profile, and the remaining contour is filled to obtain a new wire-only binarization map. ;
  • wire-only binarization map is subjected to bone refinement to obtain a skeleton of the wire
  • Step 2) through the extracted bones, calculate the end points and the bifurcation points of the wire, the specific steps are: finding the wire skeleton through the wire skeleton obtained in step 1), and the prior knowledge of the end points of the wire and the bifurcation point feature. End points and bifurcation points. If there are bifurcation points, separate the wire bones from the bifurcation points and divide them into multiple line segments. If there is no branching point, the original wire bone is a line segment, and all the line segments and the line segment end points are in the bone.
  • the type in the upper layer software is a line segment, and all the line segments and the line segment end points are in the bone.
  • Step 6 Determine whether the connection between the circuit components and the wires is accurate. The specific steps are as follows:
  • the upper layer software connects the circuit components identified in step four with the wires identified in step five, and then compares with the circuit diagrams implemented by the upper layer software. If the circuit diagram is consistent, the circuit connection is considered to be accurate.
  • the color pixel threshold is 30% of the area of each of the nine squares.
  • Figure 1 is a schematic view showing the structure of an educational toy kit of the present invention.
  • a toy kit and a circuit component and a wire identification method thereof As shown in FIG. 1, a toy kit and a circuit component and a wire identification method thereof, a bottom plate 1, a circuit component 2 and a wire 3, a bottom plate 1 is placed on a plane, and circuit components 2 and wires 3 are placed on the bottom plate 3. .
  • the bottom plate is a rectangle having rounded corners, and a calibration angle is provided at four corners of the rectangle; preferably, the calibration angle is a red circular arc line.
  • FIG. 2 is a flow chart of a method of identifying circuit components and wires in an educational toy kit of the present invention.
  • a method for identifying circuit components and wires in an educational toy kit includes the following steps:
  • Step 1 Install the game program on the tablet, and then place the bottom plate on the plane to ensure that the side of the calibration angle is facing up;
  • Step 2 complete the connection between the circuit components and the wires on the bottom plate, collect the color images in real time through the rear camera of the tablet computer, and move the tablet computer to ensure that the color images collected by the rear camera have at least three calibration angles, the specific steps are :
  • Step 3 extracting a valid identification area from the color image of step 2, the specific steps are:
  • step B) scanning the four calibration angle binary images obtained in step A) to obtain the corresponding edge contour map, and filtering out the unreasonable contour according to the prior knowledge of the eccentricity and size of the edge contour;
  • Step 4 detecting circuit components located in the effective recognition area of the color image, the specific steps are:
  • the extracted region of interest image is converted from the RGB color space to the HSV color space focusing on the color representation.
  • the specific conversion formula is:
  • H is the tone value
  • S is the saturation value
  • V is the brightness value
  • max ⁇ C(R'), C(G'), C(B') ⁇ means that one pixel is in red and green in the original image.
  • the maximum pixel value of the three channels of blue, min ⁇ C(R'), C(G'), C(B') ⁇ indicates that the pixel of one pixel in the original image is the smallest in the three channels of red, green and blue.
  • Value, and the value range of H is between 0-360;
  • the color image is binarized according to the a priori threshold in the HSV space of the color of the circuit component's outer casing.
  • the specific formula is as follows:
  • B(x, y) represents the binary pixel value of the image pixel point (x, y)
  • H(x, y) S(x, y), V(x, y) respectively represent the image pixel point (x, y) the hue value, saturation value, and brightness value in the HSV color space
  • B_H(x, y), B_S(x, y), B_V(x, y) respectively indicate whether the image pixel points (x, y) are respectively In the specified H, S, and V regions, if yes, the value is 1, otherwise, the value is 0
  • H min and H max respectively indicate the a priori color of the color of a component shell in the HSV color space.
  • S min and S max respectively represent the a priori minimum and maximum values of the saturation of the color of a component shell in the HSV color space
  • V min and V max respectively indicate the color of a component shell in the HSV.
  • the binarized image can be regarded as only two grayscale images.
  • the edge of the image refers to the part of the grayscale image where the grayscale changes are more intense.
  • the degree of change of the grayscale value is quantified by the gradient change between adjacent pixels.
  • the gradient is a two-dimensional equivalent of the first-order two-dimensional derivative.
  • G x represents the difference of adjacent pixels in the x direction
  • G y represents the difference of adjacent pixels in the y direction
  • f[i, j+1] represents the pixel value of the image in the i th row and j+1th column.
  • f[i,j] represents the pixel value of the image in the i-th row and the j-th column
  • f[i+1,j] represents the pixel value of the image in the i-th row and the j-th column
  • G(x, y) represents the gradient value at the (x, y) point of the image
  • the gradient magnitude of the edge point is calculated, and the gradient magnitude set of all the edge points is the extracted edge contour;
  • the method of calculating the non-stationary electric fan and the remaining circuit component ears is based on the prior knowledge of the shape, size and eccentricity of the circuit component's ear contour, filtering out the unreasonable contour, thereby obtaining the two ears of the circuit component. s position;
  • step b) needs to be color-divided to segment the blue binary image and the green binary image;
  • the blue binary image is scanned, and the prior knowledge of the contour shape, the eccentricity, and the size of the blade is used to determine whether there is a blade, and if so, the circuit component is determined to be a fan; otherwise, Skip directly to step 2;
  • the calculation method for the ear of the static fan is as follows: clustering the green binary image in step 1 to cluster the green dots closer to each other, through the same class
  • the number of point sets, the size of the minimum outsourcing rectangle of the point set, the eccentricity of the point set, and the prior knowledge of the point set position are used to filter the set of points that are clustered into one class. If only two types of point sets remain after filtering, the current fan blade is considered
  • the position has a large probability to cover the ear of the electrical component, and the center point position of the ear of the circuit component is calculated by the center point position of the two types of point sets and the contour information of the component;
  • the rotation angle of the circuit component is calculated by ensuring that the ear of the circuit component is in the horizontal direction. According to the calculated ear position and center position of the circuit component, to ensure that the ear of the circuit component is in the horizontal direction, the rotation angle of the circuit component is calculated;
  • the private printing color in the upper middle, middle middle and lower middle red is 1, yellow is 2, blue is 3, green is 4, no is 0, and the current color is ignored.
  • 0, 11 circuit components can be encoded, according to the circuit component code, the type of the current detection circuit component can be uniquely determined;
  • circuit component codes are as follows:
  • Step 5 detecting the wires located in the effective recognition area of the color image, the specific steps are:
  • Step 1) by dividing the wire and extracting the bone by color difference, the specific steps are as follows:
  • the image is binarized according to the a priori threshold value of the color of the wire in the HSV space to obtain a wire binarized image
  • the wire binarized image is scanned, and the contour of the wire profile is filtered by a priori knowledge of the shape and size of the wire profile, and the remaining contour is filled to obtain a new wire-only binarization map. ;
  • wire-only binarization map is subjected to bone refinement to obtain a skeleton of the wire
  • Step 2) through the extracted bones, calculate the end points and the bifurcation points of the wire, the specific steps are: finding the wire skeleton through the wire skeleton obtained in step 1), and the prior knowledge of the end points of the wire and the bifurcation point feature. End points and bifurcation points. If there are bifurcation points, separate the wire bones from the bifurcation points and divide them into multiple line segments. If there is no branching point, the original wire bone is a line segment, and all the line segments and the line segment end points are in the bone.
  • the type in the upper layer software is a line segment, and all the line segments and the line segment end points are in the bone.
  • Step 6 Determine whether the connection between the circuit components and the wires is accurate. The specific steps are as follows:
  • the upper layer software connects the circuit components identified in step four with the wires identified in step five, and then compares with the circuit diagrams implemented by the upper layer software. If the circuit diagram is consistent, the circuit connection is considered to be accurate.
  • the color pixel threshold is 30% of the area of each of the nine squares.
  • the invention intelligently combines the application of computer vision pattern recognition technology with HSV color space, binarization processing and image cutting technology, can determine the type of circuit components and wires, and can judge whether the circuit connection is accurate or not, and has an operation Fast speed, accurate positioning, good hardware and software technology, smart game interaction design; beautiful and simple, faster judgment, enhance children's imagination, increase game fun, so that children can learn basic circuit knowledge, cultivate Child's interest.
  • the detection algorithm of the invention is more scientific and mature, and combines image color conversion, image convolution, image cutting, bone refinement and other algorithms, and can quickly determine the type of circuit components and wires.
  • the calculation speed of the invention is fast; each positioning detection takes about 200ms, which provides a smooth experience for the player.
  • the performance of the invention is stable, and in the case of different illumination and different tablet computers installed in the educational toy kit, the collection and test are performed on 3,000 pictures, and the false recognition rate and the missed detection rate are below 0.2%.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Quality & Reliability (AREA)
  • Image Analysis (AREA)

Abstract

一种教育玩具套件及其电路元件和电线的识别方法,包括:底板(1)、电路元器件(2)和电线(3),底板(1)放置于平面上,电路元器件(2)和电线(3)放置于底板(1)上。将电路元器件(2)和电线(3)放置于游戏底板(1)上,在平板电脑内安装游戏程序,通过平板电脑的摄像头采集放置于底板(1)上的电路元器件(2)与电线(3)图像,再基于预定义的颜色、轮廓信息、颜色编码信息,识别电路元器件(2)与电线(3),孩子将电路元器件(2)与电线(3)相连,判断出连接好的电路是否正确,增强孩子的想象力,增加游戏趣味性,使孩子可以学习到基础电路知识,培养孩子的兴趣。

Description

一种教育玩具套件及其电路元件和电线的识别方法 技术领域
本发明涉及计算机视觉检测处理技术领域,特别涉及一种教育玩具套件及其电路元件和电线的识别方法。
背景技术
现在平板电脑上有许多有趣的幼教游戏应用程序或者儿童游戏,但往往只是让小朋友在屏幕上指指画画,互动性欠缺,长时间看着屏幕容易对眼睛造成伤害,并且缺乏物理知识的教育,逻辑思维培养欠缺;并且当下一些互动性强的传统性游戏玩具已经脱离了时代的发展,形式上无法满足孩子学习、玩耍的需求,也不便于孩子和家长的互动沟通。
为了解决上述问题,计算机视觉与图像处理技术领域成功地开发了一种教育玩具套件,包括:支架、头盔探测器和底板,在平板电脑内安装游戏程序,通过平板电脑的摄像头采集放置于平面上的底板的图像。
技术问题
上述的教育玩具套件虽然解决了平板电脑中游戏的互动性欠缺的问题,但是形式仍然单一,只是单纯的互动,并没有物理知识的嵌入,无法对孩子从小培养电学知识,亦无法做到从小培养小朋友对电学的认识和兴趣。
因此,计算机视觉检测处理技术领域急需一种教育玩具套件及其电路元件和电线的识别方法,是将电路元器件和电线放置于游戏底板上,在平板电脑内安装游戏程序,通过平板电脑的摄像头采集放置于底板上的电路元器件与电线图像,再基于预定义的颜色、轮廓信息、颜色编码信息,识别电路元件与电线,孩子将电路元器件与电线相连,判断出连接好的电路是否正确,增强孩子的想象力,增加游戏趣味性,使孩子可以学习到基础电路知识,培养孩子的兴趣。
技术解决方案
本发明为了解决上述问题,提供了一种教育玩具套件及其电路元件和电线的识别方法,技术方案如下:
一种教育玩具套件,包括:底板、电路元器件和电线底板放置于平面上,电路元器件和电线放置于底板上。
优选的,在上述的一种教育玩具套件中,底板为具有圆角的矩形,在矩形的4个角上设置有校准角。
优选的,在上述的一种教育玩具套件中,校准角为红色圆弧线。
一种教育玩具套件中电路元件和电线的识别方法,包括如下步骤:
步骤一,在平板电脑中安装游戏程序,再将底板放置于平面上,保证校准角的一面朝上
步骤二,在底板上完成电路元器件与电线的连接,通过平板电脑的后置摄像头实时采集彩色图像,移动平板电脑,保证后置摄像头采集的彩色图像中至少含有3个校准角;
步骤三,从步骤二的彩色图像中提取出有效识别区域;
步骤四,检测位于彩色图像有效识别区域内的电路元器件;
步骤五,检测位于彩色图像有效识别区域内的电线;
步骤六,判断出电路元器件与电线的连接是否准确。
优选的,在上述的一种教育玩具套件中电路元件和电线的识别方法中,步骤二中后置摄像头采集的彩色图像为Ixy,Ixy=f(x,y)=(Rxy,Gxy,Bxy),其中,(x,y)表示彩色图像像素点的位置坐标,f(x,y)表示图像在像素点坐标位置处的像素值,Rxy表示图像像素点在红色通道的色彩值,Gxy表示图像像素点在绿色通道的色彩值,Bxy表示图像像素点在蓝色通道的色彩值。
优选的,在上述的一种教育玩具套件中电路元件和电线的识别方法中,步骤三中从彩色图像中提取出有效识别区域的具体步骤为:
A)根据先验知识,在步骤二的彩色图像中分割出4块校准角区域,根据HSV空间内的先验阈值,将4块校准角区域图像进行二值化处理,得到4块校准角二值图;
B)扫描步骤A)中得到的4块校准角二值图,得到相应的边缘轮廓图,再根据边缘轮廓的离心率和大小的先验知识,过滤掉不合理的轮廓;
C)根据步骤B)得到的剩余边缘轮廓,计算出4个校准角的外接矩形,在识别过程中,当至少有三个角标内都有符合条件的校准角时,其外接矩形即为计算出的有效识别区域。
优选的,在上述的一种教育玩具套件中电路元件和电线的识别方法中,步骤四中检测位于彩色图像识别区域内的电路元器件的具步骤为:
1,由于每个电路元器件外壳的颜色不同,因此通过颜色差异,分割出各个电路元器件,并提取每个电路元器件外壳的内轮廓;
2,根据步骤1中提取出的电路元器件外壳的内轮廓,计算出每个电路元器件的位置和偏转角度;
3,依据步骤2计算出的偏转角度旋转电路元器件,再分割电路元器件,通过颜色编码识别出电路元器件的类别。
优选的,在上述的一种教育玩具套件中电路元件和电线的识别方法中,步骤1中提取每个电路元器件外壳的内轮廓的具体步骤为:
a)因为电路元器件的颜色在RGB颜色空间内不利于分割开来,对光照变化也比较敏感,所以,将提取出来的感兴趣区域图像由RGB颜色空间转换到侧重于色彩表示的HSV颜色空间,具体转换公式为:
Figure PCTCN2016105741-appb-000001
Figure PCTCN2016105741-appb-000002
其中,H表示色调值,S表示饱和度值,V表示亮度值,max{C(R′)、C(G′)、C(B′)}表示在原始图像中一个像素点在红、绿、蓝三个通道的像素最大值,min{C(R′)、C(G′)、C(B′)}表示在原始图像中一个像素点在红、绿、蓝三个通道的像素最小值,并且H的取值范围位于0-360之间;
b)在HSV颜色空间内,根据电路元器件所涉及到的颜色在HSV空间内的先验阈值,将彩色图像进行二值化处理,具体公式如下:
Figure PCTCN2016105741-appb-000003
Figure PCTCN2016105741-appb-000004
Figure PCTCN2016105741-appb-000005
在二进制图像中B(x,y)=B_H(x,y)&B_S(x,y)&B_V(x,y)时,即为生成二进制图像;
其中,B(x,y)表示图像像素点(x,y)的二进制像素值,H(x,y)、S(x,y)、V(x,y)分别表示图像像素点(x,y)在HSV颜色空间内的色调值、饱和度值、亮度值;B_H(x,y)、B_S(x,y)、B_V(x,y)分别表示图像像素点(x,y)是否分别在指定的H、S、V区域内,如果是,则取值为1,否则,取值为0;Hmin、Hmax分别表示某个元器件外壳的颜色在HSV颜色空间内色调的先验最小和最大值;Smin、Smax分别表示某个元器件外壳的颜色在HSV颜色空间内饱和度的先验最小和最大值;Vmin、Vmax分别表示某个元器件外壳的颜色在HSV颜色空间内亮度的先验最小和最大值。
c)扫描二值化图像,找出所有边缘轮廓;
二值化图像可以看作是只有两个值的灰度图像,图像的边缘是指灰度图像中灰度变化比较剧烈的部分,灰度值的变化程度采用相邻像素间的梯度变化来定量表示,梯度是一阶二维导数的二维等效式,具体计算过程为:
首先,计算相邻像素的差分,具体公式为:
Gx=f[i,j+1]-f[i,j]
Gy=f[i,j]-f[i+1,j]
其中,Gx表示相邻像素在x方向上的差分,Gy表示相邻像素在y方向上的差分,f[i,j+1]表示图像在第i行第j+1列的像素值,f[i,j]表示图像在第i行第j列的像素值;f[i+1,j]表示图像在第i+1行第j列的像素值;
进一步地,计算相邻像素间的梯度,具体公式为:
Figure PCTCN2016105741-appb-000006
其中,G(x,y)表示表示图像的在(x,y)点上梯度值,
Figure PCTCN2016105741-appb-000007
表示像素值在x方向上求导,
Figure PCTCN2016105741-appb-000008
表示像素值在y方向上求导;
进一步地,计算边缘点的梯度幅值,所有边缘点的梯度幅值集合即为提取的边缘轮廓;
进一步地,计算非静止电风扇和其余电路元器件耳朵的方法是根据电路元器件耳朵轮廓形状、大小和离心率的先验知识,滤掉不合理的轮廓,从而获得电路元器件的2个耳朵的位置;
由于静止电扇的耳朵有可能被扇叶遮住部分或者全部,因此需要对步骤b)进行颜色分割,分割出蓝色二值图和绿色二值图;
进一步地,对蓝色二值图进行扫描,通过扇叶的轮廓形状、离心率、大小的先验知识,判定是否有扇叶的存在,如果有,则判定该电路元器件为风扇;否则,直接跳到步骤2;
进一步地,判定该电路元器件为风扇后,对于静止风扇的耳朵的计算方法如下:对步骤1中绿色二值图进行聚类处理,将距离较近的绿色点聚成一类,通过同一类的点集数量、点集最小外包矩形的大小、离心率、点集位置的先验知识,对聚成一类的点集进行过滤,如果过滤后只剩下两类点集,则认为当前扇叶的位置有较大概率地遮住了电气元器件的耳朵,则通过两类点集的中心点位置和元器件的轮廓信息计算出电路元器件耳朵的中心点位置。
优选的,在上述的一种教育玩具套件中电路元件和电线的识别方法中,步骤2计算每个电路元器件的位置和偏转角度的具体步骤为:
根据计算出的电路元器件的耳朵位置、中心点位置,以保证电路元器件的耳朵在水平方向为标准,计算得出电路元器件的旋转角度。
优选的,在上述的一种教育玩具套件中电路元件和电线的识别方法中,步骤3中识别出电路元器件类别的具体步骤为:
首先,需要预先为所有电路元器件设置一种编码规则,使每个电路元器件都有唯一的编码;因为待识别电路元器件数量是有限的,故选取红、黄、蓝、绿四种易区分的颜色作为编码特征色;电路元器件的私印主要集中在上中,中中,下中三个地方,当这三个地方的某一颜色像素超过颜色像素阈值,则认为此颜色是该区域的颜色;
根据电路元器件外壳颜色,上中、中中和下中的私印颜色,红记为1,黄记为2,蓝记为3,绿记为4,无记为0,忽略当前颜色也记为0,可以将11个电路元器件进行编码,根据电路元器件编码,即可唯一确定当前检测电路元器件的类型;
过滤掉不符合编码的电路元器件,将剩余的电路元器件类型、电路元器件中心点、旋转角度一起传递给上层软件,具体电路元器件编码如下表所示:
Figure PCTCN2016105741-appb-000009
Figure PCTCN2016105741-appb-000010
优选的,在上述的一种教育玩具套件中电路元件和电线的识别方法中,颜色像素阈值为九宫格中每个宫格面积的30%。
优选的,在上述的一种教育玩具套件中电路元件和电线的识别方法中,步骤五中检测位于彩色图像有效识别区域内电线的具体步骤为:
步骤1),通过颜色差异分割出电线、提取骨骼;
步骤2),通过提取到的骨骼,计算电线的端点和分岔点。
优选的,在上述的一种教育玩具套件中电路元件和电线的识别方法中,步骤1)中通过颜色差异分割出电线、提取骨骼的具体步骤为:
首先,在步骤三的有效识别区域HSV空间内,根据电线的颜色在HSV空间内的先验阈值,将图像二值化处理,得到电线二值化图像;
进一步地,对电线二值化图像进行扫描,通过电线轮廓的形状、大小的先验知识滤除不符合电线特征的轮廓,对剩下的轮廓进行填充,得到新的只有电线的二值化图;
进一步地,对上述只有电线的二值化图进行骨骼细化,得到电线的骨骼。
优选的,在上述的一种教育玩具套件中电路元件和电线的识别方法中,步骤2)中通过提取到的骨骼,计算电线的端点和分岔点的具体步骤为:通过步骤1)中得到的电线骨骼,以及电线的端点、分岔点特征的先验知识,找到电线骨骼中的端点和分岔点,如果有分岔点,将电线骨骼从分岔点分开,分成多个线段,如果没有分岔点,原电线骨骼就是一个线段,将所有的线段以及线段端点在骨骼中的类型提供给上层软件。
优选的,在上述的一种教育玩具套件中电路元件和电线的识别方法中,步骤六中判断出电路元器件与电线的连接是否准确的具体步骤为:上层软件将步骤四识别出的电路元器件与步骤五识别出的电线连接在一起,然后与上层软件实现存储的电路图进行比较,如果与电路图一致,则认为电路连接准确。
有益效果
1、本发明巧妙的将应用计算机视觉图形识别技术与HSV颜色空间、二值化处理、图像切割技术相结合使用,能够判断出电路元器件的类型以及电线,并且能够判断出电路连接是否准确,具有运算速度快,定位准确,将硬件与软件技术很好地统一起来,游戏交互设计巧妙;美观简单,判断更加快速,增强孩子的想象力,增加游戏趣味性,使孩子可以学习到基础电路知识,培养孩子的兴趣。
2、本发明检测算法更加科学、成熟,将图像的色彩转换、图像卷积、图像切割、骨骼细化等算法相结合使用,能够快速的判断出电路元器件的类型以及电线。
3、本发明计算速度快;每次定位检测耗时在200ms左右,为玩家提供流畅的使用体验。
4、本发明性能稳定,在不同光照、对不同平板电脑安装于教育玩具套件内的情况下,针对3千幅图片进行了采集测试,误识别率和漏检率在0.2%以下。
附图说明
下面结合附图和具体实施方式来详细说明本发明:
图1是本发明一种教育玩具套件的结构示意图。
图2是本发明一种教育玩具套件中电路元件和电线的识别方法的流程图。
其中,图1-2中的附图标记与部件名称之间的对应关系为:
底板1,电路元器件2,电线3。
本发明的最佳实施方式
如图2所示,一种教育玩具套件中电路元件和电线的识别方法,包括如下步骤:
步骤一,在平板电脑中安装游戏程序,再将底板放置于平面上,保证校准角的一面朝上;
步骤二,在底板上完成电路元器件与电线的连接,通过平板电脑的后置摄像头实时采集彩色图像,移动平板电脑,保证后置摄像头采集的彩色图像中至少含有3个校准角,具体步骤为:
后置摄像头采集的彩色图像为Ixy,Ixy=f(x,y)=(Rxy,Gxy,Bxy),其中,(x,y)表示彩色图像像素点的位置坐标,f(x,y)表示图像在像素点坐标位置处的像素值,Rxy表示图像像素点在红色通道的色彩值,Gxy表示图像像素点在绿色通道的色彩值,Bxy表示图像像素点在蓝色通道的色彩值;
步骤三,从步骤二的彩色图像中提取出有效识别区域,具体步骤为:
A)根据先验知识,在步骤二的彩色图像中分割出4块校准角区域,根据HSV空间内的先验阈值,将4块校准角区域图像进行二值化处理,得到4块校准角二值图;
B)扫描步骤A)中得到的4块校准角二值图,得到相应的边缘轮廓图,再根据边缘轮廓的离心率和大小的先验知识,过滤掉不合理的轮廓;
C)根据步骤B)得到的剩余边缘轮廓,计算出4个校准角的外接矩形,在识别过程中,当至少有三个角标内都有符合条件的校准角时,其外接矩形即为计算出的有效识别区域;
步骤四,检测位于彩色图像有效识别区域内的电路元器件,具体步骤为:
1,由于每个电路元器件外壳的颜色不同,因此通过颜色差异,分割出各个电路元器件,并提取每个电路元器件外壳的内轮廓,具体步骤为:
a)因为电路元器件的颜色在RGB颜色空间内不利于分割开来,对光照变化也比较敏感,所以,将提取出来的感兴趣区域图像由RGB颜色空间转换到侧重于色彩表示的HSV颜色空间,具体转换公式为:
Figure PCTCN2016105741-appb-000011
Figure PCTCN2016105741-appb-000012
其中,H表示色调值,S表示饱和度值,V表示亮度值,max{C(R′)、C(G′)、C(B′)}表示在原始图像中一个像素点在红、绿、蓝三个通道的像素最大值,min{C(R′)、C(G′)、C(B′)}表示在原始图像中一个像素点在红、绿、蓝三个通道的像素最小值,并且H的取值范围位于0-360之间;
b)在HSV颜色空间内,根据电路元器件的外壳所涉及到的颜色在HSV空间内的先验阈值,将彩色图像进行二值化处理,具体公式如下:
Figure PCTCN2016105741-appb-000013
Figure PCTCN2016105741-appb-000014
Figure PCTCN2016105741-appb-000015
在二进制图像中B(x,y)=B_H(x,y)&B_S(x,y)&B_V(x,y)时,即为生成二进制图像;
其中,B(x,y)表示图像像素点(x,y)的二进制像素值,H(x,y)、S(x,y)、V(x,y)分别表示图像像素点(x,y)在HSV颜色空间内的色调值、饱和度值、亮度值;B_H(x,y)、B_S(x,y)、B_V(x,y)分别表示图像像素点(x,y)是否分别在指定的H、S、V区域内,如果是,则取值为1,否则,取值为0;Hmin、Hmax分别表示某个元器件外壳的颜色在HSV颜色空间内色调的先验最小和最大值;Smin、Smax分别表示某个元器件外壳的颜色在HSV颜色空间内饱和度的先验最小和最大值;Vmin、Vmax分别表示某个元器件外壳的颜色在HSV颜色空间内亮度的先验最小和最大值;
c)扫描二值化图像,找出所有边缘轮廓;
二值化图像可以看作是只有两个值得灰度图像,图像的边缘是指灰度图像中灰度变化比较剧烈的部分,灰度值的变化程度采用相邻像素间的梯度变化来定量表示,梯度是一阶二维导数的二维等效式,具体计算过程为:
首先,计算相邻像素的差分,具体公式为:
Gx=f[i,j+1]-f[i,j]
Gy=f[i,j]-f[i+1,j]
其中,Gx表示相邻像素在x方向上的差分,Gy表示相邻像素在y方向上的差分,f[i,j+1]表示图像在第i行第j+1列的像素值,f[i,j]表示图像在第i行第j列的像素值;f[i+1,j]表示图像在第i+1行第j列的像素值,
进一步地,计算相邻像素间的梯度,具体公式为:
Figure PCTCN2016105741-appb-000016
其中,G(x,y)表示表示图像的在(x,y)点上梯度值,
Figure PCTCN2016105741-appb-000017
表示像素值在x方向上求导,
Figure PCTCN2016105741-appb-000018
表示像素值在y方向上求导;
进一步地,计算边缘点的梯度幅值,所有边缘点的梯度幅值集合即为提取的边缘轮廓;
进一步地,计算非静止电风扇和其余电路元器件耳朵的方法是根据电路元器件耳朵轮廓形状、大小和离心率的先验知识,滤掉不合理的轮廓,从而获得电路元器件的2个耳朵的位置;
由于静止电扇的耳朵有可能被扇叶遮住部分或者全部,因此需要对步骤b)进行颜色分割,分割出蓝色二值图和绿色二值图;
进一步地,对蓝色二值图进行扫描,通过扇叶的轮廓形状、离心率、大小的先验知识,判定是否有扇叶的存在,如果有,则判定该电路元器件为风扇;否则,直接跳到步骤2;
进一步地,判定该电路元器件为风扇后,对于静止风扇的耳朵的计算方法如下:对步骤1中绿色二值图进行聚类处理,将距离较近的绿色点聚成一类,通过同一类的点集数量、点集最小外包矩形的大小、离心率、点集位置的先验知识,对聚成一类的点集进行过滤,如果过滤后只剩下两类点集,则认为当前扇叶的位置有较大概率地遮住了电气元器件的耳朵,则通过两类点集的中心点位置和元器件的轮廓信息计算出电路元器件耳朵的中心点位置;
2,根据步骤1中提取出的电路元器件的轮廓,计算出每个电路元器件的位置和偏转角度,具体步骤为:
根据计算出的电路元器件的耳朵位置、中心点位置,以保证电路元器件的耳朵在水平方向为标准,计算得出电路元器件的旋转角度。根据计算出的电路元器件的耳朵位置、中心点位置,以保证电路元器件的耳朵在水平方向为标准,计算得出电路元器件的旋转角度;
3,依据步骤2计算出的偏转角度旋转电路元器件,再分割电路元器件,通过颜色编码识别出电路元器件的类别,具体步骤为:
首先,需要预先为所有电路元器件设置一种编码规则,使每个电路元器件都有唯一的编码;因为待识别电路元器件数量是有限的,故选取红、黄、蓝、绿四种易区分的颜色作为编码特征色;电路元器件的私印主要集中在上中,中中,下中三个地方,当这三个地方的某一颜色像素值超过先验阈值,则认为此颜色是该区域的颜色;
根据电路元器件外壳颜色,上中、中中和下中的私印颜色,红记为1,黄记为2,蓝记为3,绿记为4,无记为0,忽略当前颜色也记为0,可以将11个电路元器件进行编码,根据电路元器件编码,即可唯一确定当前检测电路元器件的类型;
过滤掉不符合编码的电路元器件,将剩余的电路元器件类型、电路元器件中心点、旋转角度一起传递给上层软件,具体电路元器件编码如下表所示:
Figure PCTCN2016105741-appb-000019
Figure PCTCN2016105741-appb-000020
步骤五,检测位于彩色图像有效识别区域内的电线,具体步骤为:
步骤1),通过颜色差异分割出电线、提取骨骼,具体步骤为:
首先,在步骤三的有效识别区域HSV空间内,根据电线的颜色在HSV空间内的先验阈值,将图像二值化处理,得到电线二值化图像;
进一步地,对电线二值化图像进行扫描,通过电线轮廓的形状、大小的先验知识滤除不符合电线特征的轮廓,对剩下的轮廓进行填充,得到新的只有电线的二值化图;
进一步地,对上述只有电线的二值化图进行骨骼细化,得到电线的骨骼;
步骤2),通过提取到的骨骼,计算电线的端点和分岔点,具体步骤为:通过步骤1)中得到的电线骨骼,以及电线的端点、分岔点特征的先验知识,找到电线骨骼中的端点和分岔点,如果有分岔点,将电线骨骼从分岔点分开,分成多个线段,如果没有分岔点,原电线骨骼就是一个线段,将所有的线段以及线段端点在骨骼中的类型提供给上层软件;
步骤六,判断出电路元器件与电线的连接是否准确,具体步骤为:
上层软件将步骤四识别出的电路元器件与步骤五识别出的电线连接在一起,然后与上层软件实现存储的电路图进行比较,如果与电路图一致,则认为电路连接准确。
本实施例中,颜色像素阈值为九宫格中每个宫格面积的30%。
本发明的实施方式
为了使本发明技术实现的措施、创作特征、达成目的与功效易于明白了解,下面将结合本发明实施例中的附图,对本发明实施例中的技术方案进行清楚、完整地描述,显然,所描述的实施例仅仅是本发明一部分实施例,而不是全部的实施例。基于本发明中的实施例,本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例,都属于本发明保护的范围。
图1是本发明一种教育玩具套件的结构示意图。
如图1所示,一种玩具套件及其电路元件和电线的识别方法,底板1、电路元器件2和电线3,底板1放置于平面上,电路元器件2和电线3放置于底板3上。
本实施例中,底板为具有圆角的矩形,在矩形的4个角上设置有校准角;优选的,校准角为红色圆弧线。
图2是本发明一种教育玩具套件中电路元件和电线的识别方法的流程图。
如图2所示,一种教育玩具套件中电路元件和电线的识别方法,包括如下步骤:
步骤一,在平板电脑中安装游戏程序,再将底板放置于平面上,保证校准角的一面朝上;
步骤二,在底板上完成电路元器件与电线的连接,通过平板电脑的后置摄像头实时采集彩色图像,移动平板电脑,保证后置摄像头采集的彩色图像中至少含有3个校准角,具体步骤为:
后置摄像头采集的彩色图像为Ixy,Ixy=f(x,y)=(Rxy,Gxy,Bxy),其中,(x,y)表示彩色图像像素点的位置坐标,f(x,y)表示图像在像素点坐标位置处的像素值,Rxy表示图像像素点在红色通道的色彩值,Gxy表示图像像素点在绿色通道的色彩值,Bxy表示图像像素点在蓝色通道的色彩值;
步骤三,从步骤二的彩色图像中提取出有效识别区域,具体步骤为:
A)根据先验知识,在步骤二的彩色图像中分割出4块校准角区域,根据HSV空间内的先验阈值,将4块校准角区域图像进行二值化处理,得到4块校准角二值图;
B)扫描步骤A)中得到的4块校准角二值图,得到相应的边缘轮廓图,再根据边缘轮廓的离心率和大小的先验知识,过滤掉不合理的轮廓;
C)根据步骤B)得到的剩余边缘轮廓,计算出4个校准角的外接矩形,在识别过程中,当至少有三个角标内都有符合条件的校准角时,其外接矩形即为计算出的有效识别区域;
步骤四,检测位于彩色图像有效识别区域内的电路元器件,具体步骤为:
1,由于每个电路元器件外壳的颜色不同,因此通过颜色差异,分割出各个电路元器件,并提取每个电路元器件外壳的内轮廓,具体步骤为:
a)因为电路元器件的颜色在RGB颜色空间内不利于分割开来,对光照变化也比较敏感,所以,将提取出来的感兴趣区域图像由RGB颜色空间转换到侧重于色彩表示的HSV颜色空间,具体转换公式为:
Figure PCTCN2016105741-appb-000021
Figure PCTCN2016105741-appb-000022
其中,H表示色调值,S表示饱和度值,V表示亮度值,max{C(R′)、C(G′)、C(B′)}表示在原始图像中一个像素点在红、绿、蓝三个通道的像素最大值,min{C(R′)、C(G′)、C(B′)}表示在原始图像中一个像素点在红、绿、蓝三个通道的像素最小值,并且H的取值范围位于0-360之间;
b)在HSV颜色空间内,根据电路元器件的外壳所涉及到的颜色在HSV空间内的先验阈值,将彩色图像进行二值化处理,具体公式如下:
Figure PCTCN2016105741-appb-000023
Figure PCTCN2016105741-appb-000024
Figure PCTCN2016105741-appb-000025
在二进制图像中B(x,y)=B_H(x,y)&B_S(x,y)&B_V(x,y)时,即为生成二进制图像;
其中,B(x,y)表示图像像素点(x,y)的二进制像素值,H(x,y)、S(x,y)、V(x,y)分别表示图像像素点(x,y)在HSV颜色空间内的色调值、饱和度值、亮度值;B_H(x,y)、B_S(x,y)、B_V(x,y)分别表示图像像素点(x,y)是否分别在指定的H、S、V区域内,如果是,则取值为1,否则,取值为0;Hmin、Hmax分别表示某个元器件外壳的颜色在HSV颜色空间内色调的先验最小和最大值;Smin、Smax分别表示某个元器件外壳的颜色在HSV颜色空间内饱和度的先验最小和最大值;Vmin、Vmax分别表示某个元器件外壳的颜色在HSV颜色空间内亮度的先验最小和最大值;
c)扫描二值化图像,找出所有边缘轮廓;
二值化图像可以看作是只有两个值得灰度图像,图像的边缘是指灰度图像中灰度变化比较剧烈的部分,灰度值的变化程度采用相邻像素间的梯度变化来定量表示,梯度是一阶二维导数的二维等效式,具体计算过程为:
首先,计算相邻像素的差分,具体公式为:
Gx=f[i,j+1]-f[i,j]
Gy=f[i,j]-f[i+1,j]
其中,Gx表示相邻像素在x方向上的差分,Gy表示相邻像素在y方向上的差分,f[i,j+1]表示图像在第i行第j+1列的像素值,f[i,j]表示图像在第i行第j列的像素值;f[i+1,j]表示图像在第i+1行第j列的像素值,
进一步地,计算相邻像素间的梯度,具体公式为:
Figure PCTCN2016105741-appb-000026
其中,G(x,y)表示表示图像的在(x,y)点上梯度值,
Figure PCTCN2016105741-appb-000027
表示像素值在x方向上求导,
Figure PCTCN2016105741-appb-000028
表示像素值在y方向上求导;
进一步地,计算边缘点的梯度幅值,所有边缘点的梯度幅值集合即为提取的边缘轮廓;
进一步地,计算非静止电风扇和其余电路元器件耳朵的方法是根据电路元器件耳朵轮廓形状、大小和离心率的先验知识,滤掉不合理的轮廓,从而获得电路元器件的2个耳朵的位置;
由于静止电扇的耳朵有可能被扇叶遮住部分或者全部,因此需要对步骤b)进行颜色分割,分割出蓝色二值图和绿色二值图;
进一步地,对蓝色二值图进行扫描,通过扇叶的轮廓形状、离心率、大小的先验知识,判定是否有扇叶的存在,如果有,则判定该电路元器件为风扇;否则,直接跳到步骤2;
进一步地,判定该电路元器件为风扇后,对于静止风扇的耳朵的计算方法如下:对步骤1中绿色二值图进行聚类处理,将距离较近的绿色点聚成一类,通过同一类的点集数量、点集最小外包矩形的大小、离心率、点集位置的先验知识,对聚成一类的点集进行过滤,如果过滤后只剩下两类点集,则认为当前扇叶的位置有较大概率地遮住了电气元器件的耳朵,则通过两类点集的中心点位置和元器件的轮廓信息计算出电路元器件耳朵的中心点位置;
2,根据步骤1中提取出的电路元器件的轮廓,计算出每个电路元器件的位置和偏转角度,具体步骤为:
根据计算出的电路元器件的耳朵位置、中心点位置,以保证电路元器件的耳朵在水平方向为标准,计算得出电路元器件的旋转角度。根据计算出的电路元器件的耳朵位置、中心点位置,以保证电路元器件的耳朵在水平方向为标准,计算得出电路元器件的旋转角度;
3,依据步骤2计算出的偏转角度旋转电路元器件,再分割电路元器件,通过颜色编码识别出电路元器件的类别,具体步骤为:
首先,需要预先为所有电路元器件设置一种编码规则,使每个电路元器件都有唯一的编码;因为待识别电路元器件数量是有限的,故选取红、黄、蓝、绿四种易区分的颜色作为编码特征色;电路元器件的私印主要集中在上中,中中,下中三个地方,当这三个地方的某一颜色像素值超过先验阈值,则认为此颜色是该区域的颜色;
根据电路元器件外壳颜色,上中、中中和下中的私印颜色,红记为1,黄记为2,蓝记为3,绿记为4,无记为0,忽略当前颜色也记为0,可以将11个电路元器件进行编码,根据电路元器件编码,即可唯一确定当前检测电路元器件的类型;
过滤掉不符合编码的电路元器件,将剩余的电路元器件类型、电路元器件中心点、旋转角度一起传递给上层软件,具体电路元器件编码如下表所示:
Figure PCTCN2016105741-appb-000029
Figure PCTCN2016105741-appb-000030
步骤五,检测位于彩色图像有效识别区域内的电线,具体步骤为:
步骤1),通过颜色差异分割出电线、提取骨骼,具体步骤为:
首先,在步骤三的有效识别区域HSV空间内,根据电线的颜色在HSV空间内的先验阈值,将图像二值化处理,得到电线二值化图像;
进一步地,对电线二值化图像进行扫描,通过电线轮廓的形状、大小的先验知识滤除不符合电线特征的轮廓,对剩下的轮廓进行填充,得到新的只有电线的二值化图;
进一步地,对上述只有电线的二值化图进行骨骼细化,得到电线的骨骼;
步骤2),通过提取到的骨骼,计算电线的端点和分岔点,具体步骤为:通过步骤1)中得到的电线骨骼,以及电线的端点、分岔点特征的先验知识,找到电线骨骼中的端点和分岔点,如果有分岔点,将电线骨骼从分岔点分开,分成多个线段,如果没有分岔点,原电线骨骼就是一个线段,将所有的线段以及线段端点在骨骼中的类型提供给上层软件;
步骤六,判断出电路元器件与电线的连接是否准确,具体步骤为:
上层软件将步骤四识别出的电路元器件与步骤五识别出的电线连接在一起,然后与上层软件实现存储的电路图进行比较,如果与电路图一致,则认为电路连接准确。
本实施例中,颜色像素阈值为九宫格中每个宫格面积的30%。
本发明巧妙的将应用计算机视觉图形识别技术与HSV颜色空间、二值化处理、图像切割技术相结合使用,能够判断出电路元器件的类型以及电线,并且能够判断出电路连接是否准确,具有运算速度快,定位准确,将硬件与软件技术很好地统一起来,游戏交互设计巧妙;美观简单,判断更加快速,增强孩子的想象力,增加游戏趣味性,使孩子可以学习到基础电路知识,培养孩子的兴趣。
本发明检测算法更加科学、成熟,将图像的色彩转换、图像卷积、图像切割、骨骼细化等算法相结合使用,能够快速的判断出电路元器件的类型以及电线。
本发明计算速度快;每次定位检测耗时在200ms左右,为玩家提供流畅的使用体验。
本发明性能稳定,在不同光照、对不同平板电脑安装于教育玩具套件内的情况下,针对3千幅图片进行了采集测试,误识别率和漏检率在0.2%以下。
以上显示和描述了本发明的基本原理、主要特征和本发明的优点。本行业的技术人员应该了解,本发明不受上述实施例的限制,上述实施例和说明书中描述的只是说明本发明的原理,在不脱离本发明精神和范围的前提下本发明还会有各种变化和改进,这些变化和改进都落入要求保护的本发明范围内。本发明要求保护范围由所附的权利要求书及其等同物界定。
工业实用性
所属领域技术人员根据上文的记载容易得知,本发明技术方案适合在工业中制造并在生产、生活中使用,因此本发明具备工业实用性。

Claims (10)

  1. 一种教育玩具套件,其特征在于,包括:底板、电路元器件和电线,所述底板放置于平面上,所述电路元器件和电线放置于底板上。
  2. 根据权利要求1所述的一种教育玩具套件,其特征在于,所述底板为具有圆角的矩形,在所述矩形的4个角上设置有校准角;所述校准角为红色圆弧线。
  3. 一种教育玩具套件中电路元件和电线的识别方法,其特征在于,包括如下步骤:
    步骤一,在平板电脑中安装游戏程序,再将底板放置于平面上,保证校准角的一面朝上;
    步骤二,在底板上完成电路元器件与电线的连接,通过平板电脑的后置摄像头实时采集彩色图像,移动平板电脑,保证后置摄像头采集的彩色图像中至少含有3个校准角,具体步骤为:
    后置摄像头采集的彩色图像为Ixy,Ixy=f(x,y)=(Rxy,Gxy,Bxy),其中,(x,y)表示彩色图像像素点的位置坐标,f(x,y)表示图像在像素点坐标位置处的像素值,Rxy表示图像像素点在红色通道的色彩值,Gxy表示图像像素点在绿色通道的色彩值,Bxy表示图像像素点在蓝色通道的色彩值。
    步骤三,从所述步骤二的彩色图像中提取出有效识别区域;
    步骤四,检测位于彩色图像有效识别区域内的电路元器件;
    步骤五,检测位于彩色图像有效识别区域内的电线;
    步骤六,判断出电路元器件与电线的连接是否准确。
  4. 根据权利要求3所述的一种教育玩具套件中电路元件和电线的识别方法,其特征在于,所述步骤三中从彩色图像中提取出有效识别区域的具体步骤为:
    A)根据先验知识,在步骤二的彩色图像中分割出4块校准角区域,根据HSV空间内的先验阈值,将4块校准角区域图像进行二值化处理,得到4块校准角二值图;
    B)扫描步骤A)中得到的4块校准角二值图,得到相应的边缘轮廓图,再根据边缘轮廓的离心率和大小的先验知识,过滤掉不合理的轮廓;
    C)根据步骤B)得到的剩余边缘轮廓,计算出4个校准角的外接矩形,在识别过程中,当至少有三个角标内都有符合条件的校准角时,其外接矩形即为计算出的有效识别区域。
  5. 根据权利要求3所述的一种教育玩具套件中电路元件和电线的识别方法,其特征在于,所述步骤四中检测位于彩色图像识别区域内的电路元器件的具步骤为:
    1,由于每个元器件外壳的颜色不同,因此通过颜色差异,分割出各个电路元器件,并提取每个电路元器件外壳的内轮廓;
    2,根据步骤1中提取出的电路元器件外壳的内轮廓,计算出每个电路元器件的位置和偏转角度;
    3,依据步骤2计算出的偏转角度旋转电路元器件,再分割电路元器件,通过颜色编码识别出电路元器件的类别。
  6. 根据权利要求5所述的一种教育玩具套件中电路元件和电线的识别方法,其特征在于,所述步骤1中提取每个电路元器件轮廓的具体步骤为:
    a)因为电路元器件的颜色在RGB颜色空间内不利于分割开来,对光照变化也比较敏感,所以,将提取出来的感兴趣区域图像由RGB颜色空间转换到侧重于色彩表示的HSV颜色空间,具体转换公式为:
    V=max{C(R′)、C(G′)、C(B′)};
    Figure PCTCN2016105741-appb-100001
    Figure PCTCN2016105741-appb-100002
    其中,H表示色调值,S表示饱和度值,V表示亮度值,max{C(R′)、C(G′)、C(B′)}表示在原始图像中一个像素点在红、绿、蓝三个通道的像素最大值,min{C(R′)、C(G′)、C(B′)}表示在原始图像中一个像素点在红、绿、蓝三个通道的像素最小值,并且H的取值范围位于0-360之间;
    b)在HSV颜色空间内,根据电路元器件的外壳所涉及到的颜色在HSV空间内的先验阈值,将彩色图像进行二值化处理,具体公式如下:
    Figure PCTCN2016105741-appb-100003
    Figure PCTCN2016105741-appb-100004
    Figure PCTCN2016105741-appb-100005
    在二进制图像中B(x,y)=B_H(x,y)&B_S(x,y)&B_V(x,y)时,即为生成二进制图像;
    其中,B(x,y)表示图像像素点(x,y)的二进制像素值,H(x,y)、S(x,y)、V(x,y)分别表示图像像素点(x,y)在HSV颜色空间内的色调值、饱和度值、亮度值;B_H(x,y)、B_S(x,y)、B_V(x,y)分别表示图像像素点(x,y)是否分别在指定的H、S、V区域内,如果是,则取值为1,否则,取值为0;Hmin、Hmax分别表示某个元器件外壳的颜色在HSV颜色空间内色调的先验最小和最大值;Smin、Smax分别表示某个元器件外壳的颜色在HSV颜色空间内饱和度的先验最小和最大值;Vmin、Vmax分别表示某个元器件外壳的颜色在HSV颜色空间内亮度的先验最小和最大值;
    c)扫描二值化图像,找出所有边缘轮廓;
    二值化图像可以看作是只有两个值的灰度图像,图像的边缘是指灰度图像中灰度变化比较剧烈的部分,灰度值的变化程度采用相邻像素间的梯度变化来定量表示,梯度是一阶二维导数的二维等效式,具体计算过程为:
    首先,计算相邻像素的差分,具体公式为:
    Gx=f[i,j+1]-f[i,j]
    Gy=f[i,j]-f[i+1,j]
    其中,Gx表示相邻像素在x方向上的差分,Gy表示相邻像素在y方向上的差分,f[i,j+1]表示图像在第i行第j+1列的像素值,f[i,j]表示图像在第i行第j列的像素值;f[i+1,j]表示图像在第i+1行第j列的像素值,
    进一步地,计算相邻像素间的梯度,具体公式为:
    Figure PCTCN2016105741-appb-100006
    其中,G(x,y)表示表示图像的在(x,y)点上梯度值,
    Figure PCTCN2016105741-appb-100007
    表示像素值在x方向上求导,
    Figure PCTCN2016105741-appb-100008
    表示像素值在y方向上求导;
    进一步地,计算边缘点的梯度幅值,所有边缘点的梯度幅值集合即为提取的边缘轮廓;
    进一步地,计算非静止电风扇和其余电路元器件耳朵的方法是根据电路元器件耳朵轮廓形状、大小和离心率的先验知识,滤掉不合理的轮廓,从而获得电路元器件的2个耳朵的位置;
    由于静止电扇的耳朵有可能被扇叶遮住部分或者全部,因此需要对步骤b)进行颜色分割,分割出蓝色二值图和绿色二值图;
    进一步地,对蓝色二值图进行扫描,通过扇叶的轮廓形状、离心率、大小的先验知识,判定是否有扇叶的存在,如果有,则判定该电路元器件为风扇;否则,直接跳到步骤2;
    进一步地,判定该电路元器件为风扇后,对于静止风扇的耳朵的计算方法如下:对步骤1中绿色二值图进行聚类处理,将距离较近的绿色点聚成一类,通过同一类的点集数量、点集最小外包矩形的大小、离心率、点集位置的先验知识,对聚成一类的点集进行过滤,如果过滤后只剩下两类点集,则认为当前扇叶的位置有较大概率地遮住了电气元器件的耳朵,则通过两类点集的中心点位置和元器件的轮廓信息计算出电路元器件耳朵的中心点位置。
  7. 根据权利要求6所述的一种教育玩具套件中电路元件和电线的识别方法,其特征在于,所述步骤2计算每个电路元器件的位置和偏转角度的具体步骤为:
    根据计算出的电路元器件的耳朵位置、中心点位置,以保证电路元器件的耳朵在水平方向为标准,计算得出电路元器件的旋转角度。
  8. 根据权利要求5所述的一种教育玩具套件中电路元件和电线的识别方法,其特征在于,所述步骤3中识别出电路元器件类别的具体步骤为:
    首先,需要预先为所有电路元器件设置一种编码规则,使每个电路元器件都有唯一的编码;因为待识别电路元器件数量是有限的,故选取红、黄、蓝、绿四种易区分的颜色作为编码特征色;电路元器件的私印主要集中在上中,中中,下中三个地方,当这三个地方的某一颜色的像素值超过先验阈值,则认为此颜色是该区域的颜色。
    根据电路元器件外壳颜色,上中、中中和下中的私印颜色,红记为1,黄记为2,蓝记为3,绿记为4,无记为0,忽略当前颜色也记为0,可以将11个电路元器件进行编码,根据电路元器件编码,即可唯一确定当前检测电路元器件的类型;
    过滤掉不符合编码的电路元器件,将剩余的电路元器件类型、电路元器件中心点、旋转角度一起传递给上层软件,具体电路元器件编码如下表所示:
    Figure PCTCN2016105741-appb-100009
    Figure PCTCN2016105741-appb-100010
  9. 根据权利要求4所述的一种教育玩具套件中电路元件和电线的识别方法,其特征在于,所述步骤五中检测位于彩色图像有效识别区域内电线的具体步骤为:
    步骤1),通过颜色差异分割出电线、提取骨骼,具体步骤为:
    首先,在所述步骤三的有效识别区域HSV空间内,根据电线的颜色在HSV空间内的先验阈值,将图像二值化处理,得到电线二值化图像;
    进一步地,对电线二值化图像进行扫描,通过电线轮廓的形状、大小的先验知识滤除不符合电线特征的轮廓,对剩下的轮廓进行填充,得到新的只有电线的二值化图;
    进一步地,对上述只有电线的二值化图进行骨骼细化,得到电线的骨骼。
    步骤2),通过提取到的骨骼,计算电线的端点和分岔点,具体步骤为:
    通过步骤1)中得到的电线骨骼,以及电线的端点、分岔点特征的先验知识,找到电线骨骼中的端点和分岔点,如果有分岔点,将电线骨骼从分岔点分开,分成多个线段,如果没有分岔点,原电线骨骼就是一个线段,将所有的线段以及线段端点在骨骼中的类型提供给上层软件。
  10. 根据权利要求8、9所述的一种教育玩具套件中电路元件和电线的识别方法,其特征在于,所述步骤六中判断出电路元器件与电线的连接是否准确的具体步骤为:上层软件将步骤四识别出的电路元器件与步骤五识别出的电线连接在一起,然后与上层软件实现存储的电路图进行比较,如果与电路图一致,则认为电路连接准确。
PCT/CN2016/105741 2016-08-19 2016-11-14 一种教育玩具套件及其电路元件和电线的识别方法 WO2018032631A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201610698013.X 2016-08-19
CN201610698013.XA CN106355592B (zh) 2016-08-19 2016-08-19 一种教育玩具套件及其电路元件和电线的识别方法

Publications (1)

Publication Number Publication Date
WO2018032631A1 true WO2018032631A1 (zh) 2018-02-22

Family

ID=57845107

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/105741 WO2018032631A1 (zh) 2016-08-19 2016-11-14 一种教育玩具套件及其电路元件和电线的识别方法

Country Status (2)

Country Link
CN (1) CN106355592B (zh)
WO (1) WO2018032631A1 (zh)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110263608A (zh) * 2019-01-25 2019-09-20 天津职业技术师范大学(中国职业培训指导教师进修中心) 基于图像特征空间变阈值度量的电子元器件自动识别方法
CN114677586A (zh) * 2022-03-15 2022-06-28 南京邮电大学 一种物理电路实验自动识别方法
CN114882520A (zh) * 2022-07-08 2022-08-09 成都西交智汇大数据科技有限公司 一种检测电路图的方法、系统、设备及可读存储介质

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106951836B (zh) * 2017-03-05 2019-12-13 北京工业大学 基于先验阈值优化卷积神经网络的作物覆盖度提取方法
CN107590838B (zh) * 2017-08-18 2021-08-17 陕西维视智造科技股份有限公司 一种金属表面颜色视觉检测系统
CN107526888B (zh) * 2017-08-22 2024-02-20 珠海泓芯科技有限公司 电路拓扑结构的生成方法及生成装置
CN108271765B (zh) * 2018-01-05 2021-01-12 湘潭大学 一种多功能爪头监控环境机器人及其植物识别方法

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110081081A1 (en) * 2009-10-05 2011-04-07 Smith Gregory C Method for recognizing objects in images
CN104680519A (zh) * 2015-02-06 2015-06-03 四川长虹电器股份有限公司 基于轮廓和颜色的七巧板识别方法
CN105194884A (zh) * 2015-10-27 2015-12-30 上海葡萄纬度科技有限公司 教育玩具套件
CN205164140U (zh) * 2015-10-27 2016-04-20 上海葡萄纬度科技有限公司 教育玩具套件
CN105513086A (zh) * 2016-01-26 2016-04-20 上海葡萄纬度科技有限公司 一种教育玩具套件及其基于形状匹配的魔方检测定位方法
CN105719307A (zh) * 2016-01-26 2016-06-29 上海葡萄纬度科技有限公司 一种教育玩具套件及检测七巧板摆放形状、位置的方法
CN105719318A (zh) * 2016-01-26 2016-06-29 上海葡萄纬度科技有限公司 一种教育玩具套件及其基于hsv的魔方颜色识别方法

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN2428833Y (zh) * 2000-06-12 2001-05-02 南京玻璃纤维研究设计院 显示屏图象校准夹具
US7878891B2 (en) * 2007-01-29 2011-02-01 Fuji Xerox Co., Ltd. Generating polyomino video game pieces and puzzle pieces from digital photos to create photominoes
CN102049137A (zh) * 2009-11-09 2011-05-11 季春香 组合块式电子元件
CN204655822U (zh) * 2015-02-27 2015-09-23 苏州雷泰医疗科技有限公司 一种复合型质量保证模体
CN105498200A (zh) * 2016-01-26 2016-04-20 上海葡萄纬度科技有限公司 一种教育玩具套件及其七巧板颜色识别方法
CN105852968B (zh) * 2016-04-06 2018-10-16 黄斌 一种肝胆外科手术轨迹追踪装置

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110081081A1 (en) * 2009-10-05 2011-04-07 Smith Gregory C Method for recognizing objects in images
CN104680519A (zh) * 2015-02-06 2015-06-03 四川长虹电器股份有限公司 基于轮廓和颜色的七巧板识别方法
CN105194884A (zh) * 2015-10-27 2015-12-30 上海葡萄纬度科技有限公司 教育玩具套件
CN205164140U (zh) * 2015-10-27 2016-04-20 上海葡萄纬度科技有限公司 教育玩具套件
CN105513086A (zh) * 2016-01-26 2016-04-20 上海葡萄纬度科技有限公司 一种教育玩具套件及其基于形状匹配的魔方检测定位方法
CN105719307A (zh) * 2016-01-26 2016-06-29 上海葡萄纬度科技有限公司 一种教育玩具套件及检测七巧板摆放形状、位置的方法
CN105719318A (zh) * 2016-01-26 2016-06-29 上海葡萄纬度科技有限公司 一种教育玩具套件及其基于hsv的魔方颜色识别方法

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110263608A (zh) * 2019-01-25 2019-09-20 天津职业技术师范大学(中国职业培训指导教师进修中心) 基于图像特征空间变阈值度量的电子元器件自动识别方法
CN110263608B (zh) * 2019-01-25 2023-07-07 天津职业技术师范大学(中国职业培训指导教师进修中心) 基于图像特征空间变阈值度量的电子元器件自动识别方法
CN114677586A (zh) * 2022-03-15 2022-06-28 南京邮电大学 一种物理电路实验自动识别方法
CN114677586B (zh) * 2022-03-15 2024-04-05 南京邮电大学 一种物理电路实验自动识别方法
CN114882520A (zh) * 2022-07-08 2022-08-09 成都西交智汇大数据科技有限公司 一种检测电路图的方法、系统、设备及可读存储介质

Also Published As

Publication number Publication date
CN106355592B (zh) 2020-06-16
CN106355592A (zh) 2017-01-25

Similar Documents

Publication Publication Date Title
WO2018032631A1 (zh) 一种教育玩具套件及其电路元件和电线的识别方法
WO2017128605A1 (zh) 一种教育玩具套件及其基于hsv的魔方颜色识别方法
WO2021138995A1 (zh) 一种棋盘格角点全自动检测方法
WO2018032630A1 (zh) 一种教育玩具套件及利用颜色和轮廓识别编程模块的方法
WO2017092431A1 (zh) 基于肤色的人手检测方法及装置
WO2018032626A1 (zh) 一种教育玩具套件及其数字识别方法
WO2017128606A1 (zh) 一种教育玩具套件及其七巧板颜色识别方法
WO2017128604A1 (zh) 一种教育玩具套件及其基于形状匹配的魔方检测定位方法
CN107154058B (zh) 一种引导使用者还原魔方的方法
KR101035768B1 (ko) 립 리딩을 위한 입술 영역 설정 방법 및 장치
CN108256467B (zh) 一种基于视觉注意机制和几何特征的交通标志检测方法
CN103218605A (zh) 一种基于积分投影与边缘检测的快速人眼定位方法
CN106203461B (zh) 一种图像处理方法及装置
WO2017128602A1 (zh) 一种教育玩具套件及其定位孔检测定位方法
CN106529520A (zh) 基于运动员号码识别的马拉松比赛照片管理方法
CN106529531A (zh) 一种基于图像处理的中国象棋识别系统及方法
CN111695373B (zh) 斑马线的定位方法、系统、介质及设备
JP2018018173A (ja) 画像処理装置、画像処理方法、コンピュータプログラム
CN108022245B (zh) 基于面线基元关联模型的光伏面板模板自动生成方法
CN110956184B (zh) 一种基于hsi-lbp特征的抽象图方向确定方法
CN107368826A (zh) 用于文本检测的方法和装置
CN108388898A (zh) 基于连接体和模板的字符识别方法
CN108682021A (zh) 快速手部跟踪方法、装置、终端及存储介质
CN108564020B (zh) 基于全景3d图像的微手势识别方法
CN113052194A (zh) 一种基于深度学习的服装色彩认知系统及其认知方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 16913386

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 16913386

Country of ref document: EP

Kind code of ref document: A1