CN-114170329-B - Color distinguishing method for positive group tobacco leaves

CN114170329BCN 114170329 BCN114170329 BCN 114170329BCN-114170329-B

Abstract

The invention discloses a color distinguishing method of positive group tobacco leaves, and relates to the technical field of tobacco leaf grading. The invention discloses a method for distinguishing the colors of positive group tobacco leaves, which comprises the steps of selecting a positive group tobacco leaf standard color sample by a tobacco leaf grading expert, obtaining a positive image of the positive group tobacco leaf standard color sample by an industrial camera, carrying out binary segmentation on the image, determining an image tobacco leaf area, sampling pixel points of the image tobacco leaf area to obtain the color difference of the tobacco leaves, wherein the positive group three-color tobacco leaf color difference value of a scatter diagram of the sample about the color difference has obvious interval distribution, taking a cross domain existing in each interval continuously as the cross domain range of the color difference interval of the tobacco leaves, defining the color difference interval of the positive group lemon yellow, orange yellow and red brown three-color tobacco leaves by calculating the threshold value of interval division through the cross domain sample point.

Inventors

ZHANG QINGGANG
ZHANG YINHUI
CHEN DONGDONG
YANG HONGKUAN
ZHANG JIWU
SUN HAOWEI
LONG JIE
ZHANG KE
ZHANG XIAOWEI
Gai Xiaolei
HE ZIFEN
HE PENGFEI

Assignees

云南省烟草质量监督检测站
云南省烟草质量监督检测站
昆明理工大学
昆明理工大学

Dates

Publication Date: 20260421
Application Date: 20211020
Priority Date: 20211020

Claims (9)

1. The color distinguishing method for the positive group tobacco leaves is characterized by comprising the following steps of: s1, manufacturing a standard color sample of the color-graded positive group tobacco leaves; S2, acquiring a front image of a standard color sample of the positive group tobacco leaves, and taking chromatic aberration as a color index; S3, sampling pixel points in the tobacco leaf area of the image at regular intervals according to pixel point coordinates to calculate pixel point chromatic aberration, and taking the average value of the sampled pixel point chromatic aberration as tobacco leaf chromatic aberration; S4, drawing a scatter diagram of a sample about chromatic aberration according to the distribution of chromatic aberration of tobacco leaves, and determining a cross domain range of a grading three-color tobacco leaf chromatic aberration taking part; And S5, calculating a threshold value of interval division by using the cross-domain sample points, and defining color difference intervals of the three-color tobacco leaves in the grades of lemon yellow, orange yellow and red brown of the positive group, wherein each interval represents different colors of the tobacco leaves in the positive group.
2. The method for distinguishing colors of positive group tobacco leaves according to claim 1, wherein the positive group tobacco leaf standard color sample is selected by a tobacco leaf grading expert, the change of the colors of the positive group tobacco leaves from light to dark is marked by using small to large Arabic numerals, a front image of the positive group tobacco leaf standard color sample is obtained by an industrial camera, the image is subjected to binarization segmentation to determine an image tobacco leaf area, the pixel points of the image tobacco leaf area are sampled to obtain tobacco leaf color differences, the positive group three-color tobacco leaf color difference value of a scatter diagram of the sample about the color differences has obvious interval distribution, the cross domain of the color difference interval of the tobacco leaves is regarded as the cross domain range of the cross domain, and the color difference interval of the positive group lemon yellow, orange and red-brown tobacco leaves is defined by calculating the threshold value of the interval division through the cross domain sample points, and each interval represents different colors of the positive group tobacco leaves.
3. The method of claim 2, wherein the color of the tobacco color of the "standard" color sample of the positive group of tobacco is marked as C, wherein: and when C=1, the tobacco leaf is the lightest piece of all the positive group tobacco leaves in the sample, and the larger C is, the darker the color of the tobacco leaf is.
4. The method for distinguishing colors of positive group tobacco leaves according to claim 3, wherein the image is subjected to binarization segmentation, a tobacco leaf sample image is converted from an RGB image to an HSV image, the HSV image is subjected to slicing processing to obtain an image of a chromaticity channel H, the H channel image is subjected to filtering processing, the filtered image is subjected to binarization threshold segmentation by a maximum inter-class variance method, the pixel values of the H channel image are extracted to be in obvious bimodal distribution, a tobacco leaf area subjected to color difference calculation is obtained after the binarization segmentation, and the image tobacco leaf area is: wherein: for pixel point coordinates in the binarized image, Is the pixel value of the coordinate.
5. The method for distinguishing color of positive group tobacco leaves according to claim 4, wherein the color difference of tobacco leaves in the tobacco leaf area is calculated, the pixel value of an original image is converted from an RGB value to a Lab value, coordinates of the tobacco leaf area in the binarized image are extracted, and the color corresponding to the Lab value of a pixel point under the coordinates is obtained and recorded as , The color of the selected anchor point in the lab space is recorded as Wherein: The color difference is calculated by using a CIEDE2000 color difference formula, and the pixel point is obtained by calculation The color difference between the color of the anchor point and the color of the anchor point is recorded as ; The tobacco color difference tobacco leaves are subjected to color difference by pixel points in tobacco leaf areas Averaging, the formula is: wherein n represents the number of pixels for performing color difference calculation, and n cannot exceed the total number of pixels in the tobacco leaf area at maximum.
6. The method for distinguishing colors of positive group tobacco leaves according to claim 4, wherein the color difference of tobacco leaves in the tobacco leaf area is calculated, and the color difference of tobacco leaves in the tobacco leaf area is calculated in the image Taking out , Is taken as a sampling pixel point, wherein P=5, 10, 15..50, which is a multiple of 5, i.e. the sampling area is: randomly selecting a tobacco leaf image, calculating the color difference of tobacco leaves according to different values P, calculating the average value of the color differences of 10 groups in total, selecting a group with the smallest color difference and average color difference, and taking the P value of the group as the P value used for calculating the selected sampling point.
7. The method for distinguishing colors of positive group tobacco leaves according to claim 2, wherein the drawing of the scatter diagram of the samples about the color differences is characterized in that the color differences are extracted by using positive group tobacco leaves marked with color shade change digital labels, and statistics is carried out on all the positive group tobacco leaf color differences to obtain the scatter diagram of the samples about the color differences, and the samples are ordered from large to small according to the digital labels: the sample has a scatter diagram of chromatic aberration, positive group three-color tobacco color difference values have obvious interval distribution, and cross domains existing in continuous intervals are used as cross domain ranges of tobacco color difference intervals.
8. The method for distinguishing colors of positive group tobacco leaves according to claim 7, wherein the cross-domain range of the color difference interval of tobacco leaves is set to be lemon yellow tobacco leaves color difference The color difference of the orange tobacco leaves is The red brown tobacco has the color difference of Wherein C is a digital label of which the tobacco leaves represent the color shade, and C epsilon {1,2,3.. N }; A scatter diagram of samples about chromatic aberration, wherein the abscissa is the samples, the samples are ordered from small to large according to the digital label, the ordinate is the chromatic aberration of tobacco leaves, 5 points are selected from a first lemon yellow tobacco sample point, and if other orange yellow tobacco sample points exist besides the adjacent orange yellow tobacco sample points, the chromatic aberration of the sample points is defined For the lower limit of the color difference crossing area of lemon yellow and orange yellow, if the sample point meets the condition, starting from the sample point forward to obtain the color difference of the sample point of lemon yellow tobacco with the largest encountered color difference value And starting from a first orange tobacco sample point with a color difference value larger than that of the orange and lemon color difference crossing areas, proceeding along the abscissa direction until encountering a first reddish brown sample point, and determining the lower limit and the upper limit of the color difference value of the orange and reddish brown color difference crossing areas by adopting the same method.
9. The method for distinguishing colors of positive group tobacco leaves according to claim 8, wherein the average value of color differences of all tobacco leaf sample points in the cross domain is used as a threshold value for distinguishing two colors of tobacco leaves: The color difference intersection area refers to a lemon yellow color difference intersection area and a orange yellow color difference intersection area, so that the lemon yellow color difference and the orange yellow color difference are respectively distinguished by the average value of the color differences of all tobacco sample points in the lemon yellow color difference intersection area and the orange yellow color difference intersection area, and the orange yellow color difference and the red brown color difference are respectively distinguished by the average value of the color differences of all tobacco sample points in the orange yellow color difference intersection area and the red brown color difference intersection area.

Description

Color distinguishing method for positive group tobacco leaves Technical Field The invention relates to a color distinguishing method of positive group tobacco leaves, and relates to the technical field of tobacco leaf grading. Background Tobacco is a raw material for producing tobacco products, the quality of the tobacco has direct and important influence on the quality of finished cigarettes, in order to strengthen the control of the quality of the tobacco, the development of the whole tobacco industry is promoted, national standards related to tobacco grading are formulated in China, tobacco grading work is carried out for a long time by tobacco grading staff in combination with tobacco grading national standards, actual conditions of tobacco producing areas and personal grading experience, obvious subjectivity is provided, and meanwhile, a great deal of manpower and financial loss exists, the standards are not uniform, and the distinction is large. The color of flue-cured tobacco refers to the state of the relevant colors, color saturation and color values of the tobacco after the same type of tobacco is modulated, the difference of the colors of the tobacco reflects the difference of the proportions of various pigments in the tobacco to a great extent, the differences are closely related to the appearance quality and the internal quality of the tobacco, therefore, the color characteristics of the tobacco are one of important grouping factors of the grading standard of the flue-cured tobacco, generally, the color characteristics of the tobacco are grouped by the depth of looking at the yellow color of the tobacco by experts, the leaf color of the positive group of tobacco is divided into lemon yellow, orange yellow and reddish brown, the specific expression of the positive group of tobacco is that the lemon yellow tobacco is composed of 100% yellow, the orange yellow tobacco is composed of 70% yellow and 30% red, the reddish brown tobacco is composed of 70% red and 30% yellow, but different judges have certain deviation on the differentiation of the color characteristics of the tobacco due to the difference of individual vision, people can make very accurate differentiation on the colors of the tobacco through inputting the image learning characteristics of the tobacco by means of a deep learning network, but have the corresponding technical problems that the obtained differentiation results cannot be effectively interpreted, and the influence on the obtained differentiation result is not enough to realize the automatic grading of the digital image of the tobacco. In summary, although the deep learning network in the prior art can distinguish the colors of tobacco leaves, the interpretation of distinguishing results and distinguishing precision can not meet the requirements of digital images of tobacco leaves and automatic grading of the whole tobacco leaves. Disclosure of Invention The invention aims to provide a color distinguishing method of positive group tobacco leaves, which is characterized in that color characteristics of a manufactured positive group tobacco leaf standard color sample are extracted, the extracted color characteristics are analyzed to define a color difference interval of the positive group tobacco leaves, a describable tobacco leaf color distinguishing method is established, and the color distinguishing of the positive group tobacco leaves is completed by utilizing a single index. In order to achieve the technical purpose and the technical effect, the invention is realized by the following technical scheme: a color distinguishing method of positive group tobacco leaves comprises the following steps: s1, manufacturing a standard color sample of the color-graded positive group tobacco leaves; S2, acquiring a front image of a standard color sample of the positive group tobacco leaves, and taking chromatic aberration as a color index; S3, sampling pixel points in the tobacco leaf area of the image at regular intervals according to pixel point coordinates to calculate pixel point chromatic aberration, and taking the average value of the sampled pixel point chromatic aberration as tobacco leaf chromatic aberration; S4, drawing a scatter diagram of a sample about chromatic aberration according to the distribution of chromatic aberration of tobacco leaves, and determining a cross domain range of a grading three-color tobacco leaf chromatic aberration taking part; And S5, calculating a threshold value of interval division by using the cross-domain sample points, and defining color difference intervals of the three-color tobacco leaves in the grades of lemon yellow, orange yellow and red brown of the positive group, wherein each interval represents different colors of the tobacco leaves in the positive group. Further, the standard color samples of the positive group tobacco leaves are selected by tobacco leaf grading specialists, the light-to-dark change of the colors of the positive grou