Device for reading resistor value with computer vision and method thereof

Patent metadata

Field	Value
Publication number	US-9959616-B2
Application number	US-201615209770-A
Country	US
Kind code	B2
Filing date	Jul 14, 2016
Priority date	Feb 2, 2016
Publication date	May 1, 2018
Grant date	May 1, 2018

Abstract

Disclosed are a cost-effective light source device that can eliminate highlights or hotspots and reduce halo phenomenon on resistor surface and a method for reading resistor value with computer vision, in which an image of a body of a resistor is extracted through irradiation by an annular light source and reflection by a replaceable diffuse reflection hood such as paper box, paper bowl, or paper cup inside pasted with white paper and extraction by an image extraction device, followed by a series of image processing operations conducted by a control unit to calculate a disposition angle of the resistor in an image. Luminous elements on opposite axial ends of the resistor are activated, according to the disposition angle, for illumination, while luminous elements on opposite lateral sides of the resistor are shut down so that a clear color resistor image can be obtained.

First claim

We claim:

1. A method for reading resistor value with computer vision comprising:
acquiring an image of a resistor to extract color bands;
training the color bands for 4-band, 5-band and establishing colors of a body of the resistor in a control unit and storing different files of color classifier for each 4-band, 5-band, and resistor body color;
in reading a resistor, extracting an image of the resistor to be read;
converting the color image so extracted into a gray-scale image, executing binarization treatment, and executing noise removal;
calculating a resistor disposition angle in the image, activating luminous elements at opposite axial ends of the disposition position of the resistor according to the angle for illumination, while luminous elements on opposite lateral sides of the resistor are deactivated in order to extract a color resistor image having no light reflection and low halo;
adjusting the clear color resistor image such that the resistor in the image is set to the horizontal and separating a resistor body image;
establishing a resistor body color image diagram having a size that is the same as that of the resistor body image and carrying colors;
executing differencing between the resistor body image and the resistor body color image diagram and converting a result color image of differencing into a gray-scale image and executing binarization on the gray-scale image with a binarization process to generate a binarized image; and
calculating coordinates of a centroid of each of the color bands of the resistor in the binarized image and extracting the color bands according to an x-axis coordinate of the centroid and a y-axis coordinate of a centroid of the resistor body image.

2. The method for reading resistor value with computer vision according to claim 1 , wherein the binarization treatment comprises a first stage of detecting a directional angle of the resistor of the resistor image and a second stage of extracting the color bands.

3. The method for reading resistor value with computer vision according to claim 1 , wherein the resistor image, after binarization, is subjected to convex hull treatment to generate one or multiple blocks.

4. The method for reading resistor value with computer vision according to claim 1 , wherein adjusting the clear color resistor image to the horizontal comprises calculating a directional angle of a single block or a largest one of multiple blocks with a moment of inertia based process and calculating an angle of a centroid line of the block with a lowest moment of inertial process.

5. The method for reading resistor value with computer vision according to claim 1 , wherein a y-axis range of a resistor body is separated by re-calculating coordinates of the centroid of the clear color resistor image that has been adjusted to the horizontal are calculated and selecting 40 pixels above and 40 pixels below a y-axis coordinate of the centroid, reducing an x-axis range of the y-range resistor body by calculating coordinates of intersection points of top and bottom boundaries of the y-axis range, comparing and selecting a large one of minimums of the x-axis coordinates of the intersection points of the top and bottom boundaries as an x-axis range in a left end and comparing and selecting a small one of maximums of the x-axis coordinates of the intersection points of the top and bottom boundaries as an x-axis range in a right end so as to separate a resistor body image.

6. The method for reading resistor value with computer vision according to claim 1 , wherein the step of comparing and selecting a relatively large one of minimums of the x-axis coordinates of the intersection points of the top and bottom boundaries as an x-axis range in a left end comprises providing histograms for values of basic colors of red, green, and blue of pixels of the resistor body image, counting individual pixel number of the red values, the green values, and the blue values, and selecting the red value having a greatest pixel number, the green value having a greatest pixel number, and a blue value having a greatest pixel number to establish the resistor body color image diagram.

7. The method for reading resistor value with computer vision according to claim 1 , wherein extracting color bands comprises differencing between the resistor body image and the resistor body color image diagram and binarization such that a binarized color band image is obtained and a centroid thereof is calculated to extract a color band sample.

8. The method for reading resistor value with computer vision according to claim 1 , wherein extracting color bands is conducted in sequence in a direction from left to right such that the color bands are extracted and converted into color codes with which a predetermined formula is applied to calculate the resistor value.

9. The device and method for reading resistor value with computer vision according to claim 8 , wherein calculating the resistor value is such that positions of the color bands are analyzed and when a first one or second one of the color bands on the left end of the resistor is recognized as a gold or silver color, the sequence of reading the color codes is inversed for calculating the resistor value.

10. The method for reading resistor value with computer vision according to claim 8 further comprising sequentially determining a centroid distance between the first color band and the second color band on the left end and a centroid distance between a first color band and a second color band on a right end such that a mechanism of comparison is activated only when the first color band and the last color band are brown, red, green, blue, purple, and grey and when the centroid distance between the first color band and the second color band on the left end is greater that the centroid distance between the first color band and second color band on the right end, the sequence of reading color codes is inversed for calculating the resistor value.

11. The method for reading resistor value with computer vision according to claim 1 , wherein an application comprises an observation table on which a resistor is positionable, an annular light source, a reflection hood, a microscope, and a control unit such that in file establishing or reading of resistor colors, the resistor is positioned on the observation table and through irradiation by the annular light source and reflection by the reflection hood and extraction and transmission of a resistor image by the microscope to the control unit, file establishing or reading of a resistor value can be performed.

12. The method for reading resistor value with computer vision according to claim 3 , wherein adjusting the clear color resistor image to the horizontal comprises calculating a directional angle of a single block or a largest one of multiple blocks with a moment of inertia based process and calculating an angle of a centroid line of the block with a lowest moment of inertial process.

Assignees

• Univ Yuan Ze

Inventors

• Chen Yung-Sheng
• Wang Jeng-Yau

Classifications

• G06V10/56

  relating to colour · CPC title

• G06V10/243

  by compensating for image skew or non-uniform image deformations · CPC title

• G06V10/26

  Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion · CPC title

• G06V10/141

  Control of illumination · CPC title

• G06V10/28

  Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns · CPC title