CN108038838A - A kind of cotton fibriia species automatic testing method and system - Google Patents
A kind of cotton fibriia species automatic testing method and system Download PDFInfo
- Publication number
- CN108038838A CN108038838A CN201711078382.XA CN201711078382A CN108038838A CN 108038838 A CN108038838 A CN 108038838A CN 201711078382 A CN201711078382 A CN 201711078382A CN 108038838 A CN108038838 A CN 108038838A
- Authority
- CN
- China
- Prior art keywords
- image
- cotton
- carrying
- linen
- processing
- Prior art date
- Legal status (The legal status 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 status listed.)
- Pending
Links
- 229920000742 Cotton Polymers 0.000 title claims abstract description 72
- 238000012360 testing method Methods 0.000 title abstract description 9
- 239000000835 fiber Substances 0.000 claims abstract description 133
- 238000012545 processing Methods 0.000 claims abstract description 53
- 238000000034 method Methods 0.000 claims abstract description 51
- 238000001514 detection method Methods 0.000 claims abstract description 34
- 238000007781 pre-processing Methods 0.000 claims abstract description 17
- 238000000605 extraction Methods 0.000 claims description 25
- 238000013528 artificial neural network Methods 0.000 claims description 14
- 238000001914 filtration Methods 0.000 claims description 14
- 239000012535 impurity Substances 0.000 claims description 12
- 230000008030 elimination Effects 0.000 claims description 9
- 238000003379 elimination reaction Methods 0.000 claims description 9
- 230000001629 suppression Effects 0.000 claims description 9
- 238000003708 edge detection Methods 0.000 claims description 5
- 238000002156 mixing Methods 0.000 abstract 1
- 230000006870 function Effects 0.000 description 14
- 244000025254 Cannabis sativa Species 0.000 description 10
- 235000012766 Cannabis sativa ssp. sativa var. sativa Nutrition 0.000 description 10
- 235000012765 Cannabis sativa ssp. sativa var. spontanea Nutrition 0.000 description 10
- 235000009120 camo Nutrition 0.000 description 10
- 235000005607 chanvre indien Nutrition 0.000 description 10
- 239000011487 hemp Substances 0.000 description 10
- 240000008564 Boehmeria nivea Species 0.000 description 9
- 238000010606 normalization Methods 0.000 description 6
- 239000004753 textile Substances 0.000 description 5
- 238000012549 training Methods 0.000 description 5
- 238000012546 transfer Methods 0.000 description 4
- 230000009466 transformation Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 238000012937 correction Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000012886 linear function Methods 0.000 description 3
- 238000012935 Averaging Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 239000003086 colorant Substances 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000008451 emotion Effects 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000005286 illumination Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000008439 repair process Effects 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 208000003464 asthenopia Diseases 0.000 description 1
- 238000010219 correlation analysis Methods 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 238000011978 dissolution method Methods 0.000 description 1
- 238000005315 distribution function Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000001125 extrusion Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 230000010355 oscillation Effects 0.000 description 1
- 230000002035 prolonged effect Effects 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30124—Fabrics; Textile; Paper
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Quality & Reliability (AREA)
- Treatment Of Fiber Materials (AREA)
- Image Processing (AREA)
Abstract
The present invention relates to a kind of cotton fibriia species automatic testing method and system.The described method includes：Receive the original image of the cotton fibriia of image acquisition device；Image preprocessing is carried out to the original image；Image procossing is carried out to pretreated original image, obtains processing image；Extract the characteristic parameter of cotton fibriia described in the processing image；The kinds of fibers of the cotton fibriia is identified according to the characteristic parameter.Technical scheme can realize the quick detection to cotton fibriia species in fiber blending products, especially cotton, ramee species, improve the accuracy rate of detection, reduce testing cost.
Description
Technical Field
The invention relates to the technical field of textile fiber detection, in particular to a method and a system for automatically detecting the type of cotton and linen fibers.
Background
The export of the textile needs to be detected strictly, China is taken as the biggest textile production and export country in the world, the identification of the fiber types in the textile is mainly detected manually at present, the working efficiency is low, and the identification is easily influenced by subjective factors of people to cause wrong identification.
The identification of the type of fibres plays an extremely important role in the textile processing industry. Because many fiber blended products have similar chemical properties, the proportion of each fiber cannot be determined by a chemical dissolution method, and identification is usually carried out by a microscope method.
However, the detection by using a microscope manually requires a long time and high cost, and the result of the identification is influenced by subjective emotion and visual fatigue due to long-term work, so that the false detection rate and the false detection rate are high.
First, the testing time of conventional testing methods, such as testing cotton/hemp fiber types, is typically at least 1500 to 4000 fibers in a dark room requiring at least two testing personnel, perhaps one am or more, and sometimes only one to two specimens. Secondly, the traditional detection method has low accuracy of the detection result, and the detection result has errors due to the fact that inspectors are inevitably influenced by subjective emotion and long-time work, and in addition, when the fiber section is sliced, the existing slicing method basically uses collodion to coat fibers as an embedding medium, the embedding medium hardly meets the requirements on strength, the phenomenon of local extrusion is inevitably generated on the fibers, so that the fiber section deforms, the real fiber section shape cannot be reflected when the detection is performed, and the error of the detection of the fiber section shape is directly increased. Thirdly, the traditional detection method has higher cost and increasingly high labor cost, and in addition, the detection cost is higher when the detection period is longer or the detection is carried out again and again due to errors generated in the detection.
Disclosure of Invention
The invention provides a cotton and hemp fiber type automatic detection method and system in order to realize the rapid detection of cotton and hemp fiber types, particularly cotton and ramie fiber types in fiber blended products, improve the detection accuracy and reduce the detection cost.
In one aspect, the invention provides a method for automatically detecting the type of cotton and linen fibers, which comprises the following steps:
step 1, receiving an original image of the cotton and linen fibers acquired by an image acquisition device;
step 2, carrying out image preprocessing on the original image;
step 3, carrying out image processing on the preprocessed original image to obtain a processed image;
step 4, extracting characteristic parameters of the cotton and linen fibers in the processed image;
and 5, identifying the fiber type of the cotton and linen fibers according to the characteristic parameters.
On the other hand, the invention provides an automatic cotton and hemp fiber type detection system, which comprises a receiving unit, a preprocessing unit, a processing unit, an extraction unit and an identification unit;
the receiving unit is used for receiving the original image of the cotton and linen fibers acquired by the image acquisition device;
the preprocessing unit is used for preprocessing the original image;
the processing unit is used for carrying out image processing on the preprocessed original image to obtain a processed image;
the extraction unit is used for extracting the characteristic parameters of the cotton and linen fibers in the processed image;
the identification unit is used for identifying the fiber type of the cotton and linen fibers according to the characteristic parameters.
The method and the system for automatically detecting the cotton and linen fiber types have the advantages that the original images of the cotton and linen fibers, such as cotton and ramie fibers, are acquired and obtained through the cooperation of an image acquisition device such as an electron microscope and a proper light source. The original image is subjected to image preprocessing by a computer, including processing such as graying, enhancement, noise elimination and the like on the image so as to eliminate the influence of illumination unevenness and noise generated in the acquisition and transmission processes of the image, and the image is subjected to processing such as correction, impurity filtering, contour extraction, repair and the like so as to obtain a processed image which is beneficial to extracting characteristic parameters. And obtaining the characteristic parameters with the highest correlation with the fiber types through correlation calculation, extracting the characteristic parameters in the processed image, identifying the fiber types based on the characteristic parameters through a method such as a neural network, and counting the number of various fibers. Therefore, the rapid detection of the cotton and hemp fiber types in the fiber blended product is realized, the detection accuracy is improved, and the detection cost is reduced.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a schematic flow chart of a method for automatically detecting the type of a cotton or linen fiber according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of images at various stages of an image processing process according to an embodiment of the present invention;
fig. 3 is a block diagram of an automatic detecting system for detecting the type of cotton and linen fibers according to an embodiment of the present invention.
Detailed Description
The principles and features of this invention are described below in conjunction with the following drawings, which are set forth by way of illustration only and are not intended to limit the scope of the invention.
As shown in fig. 1, an automatic detection method for a type of a cotton and linen fiber provided by an embodiment of the present invention includes:
step 1, receiving an original image of the cotton and linen fibers acquired by an image acquisition device.
And 2, carrying out image preprocessing on the original image.
And 3, carrying out image processing on the preprocessed original image to obtain a processed image.
And 4, extracting characteristic parameters of the cotton and linen fibers in the processed image.
And 5, identifying the fiber type of the cotton and linen fibers according to the characteristic parameters.
In this embodiment, the original image of the cotton and hemp fiber, such as cotton and ramie fiber, is acquired by the image acquisition device, such as an electron microscope, and the image acquisition device is matched with a suitable light source. The computer carries out image preprocessing on the original image, including graying, enhancing, denoising and other processing on the image so as to eliminate the influence of uneven illumination and noise generated in the acquisition and transmission processes of the image, and carries out correction, impurity filtering, contour extraction, repair and other processing on the image so as to obtain a processed image beneficial to extracting characteristic parameters. And obtaining the characteristic parameters with the highest correlation with the fiber types through correlation calculation, extracting the characteristic parameters from the processed images, identifying the fiber types through a neural network method based on the characteristic parameters, and counting the number of various fibers. Therefore, the rapid detection of the cotton and hemp fiber types in the fiber blended product is realized, the detection accuracy is improved, and the detection cost is reduced.
Preferably, step 2 specifically comprises:
and 2.1, carrying out graying processing on the original image according to a weighted average value method to obtain a grayscale image.
Since the color elements of the original image do not carry necessary characteristic parameters required in the subsequent processing, the original image is first subjected to gray scale processing in order to reduce the processing load. The weighted average method has high operation efficiency and good processing result, so that gray level processing is performed on the acquired original images of the cotton and hemp fibers by adopting a weighted average value method, R, G, B (channel values of three primary colors of red, green and blue) are endowed with different weight values according to importance and other indexes, and after the weight values are endowed, R, G, B endowed with the weight values are subjected to average value processing, namely R, G, B is made to be the same through the following formula.
R＝G＝B＝(WRR+WGG+WBB)，
Wherein, WR、WG、WRRespectively R, G, B.
And 2.2, carrying out image enhancement processing on the gray level image according to a histogram enhancement method to obtain an enhanced image.
In order to enhance local features in an image, enhancement processing needs to be performed on the image. The histogram of an image is a very important statistical feature of the image, and can be regarded as an approximation of an image density function, and the image can be enhanced by improving the image histogram. Typically the grey scale density of an image is related to where the pixel is located. For example, the gray scale distribution density function at point (x, y) of the image is p (z, x, y), then the gray scale density function of the image is as follows:
wherein x and y represent coordinates of pixel points in the image, z represents a gray value, D represents a definition domain of the image, and S represents the area of the definition domain D in the image.
It is often difficult to accurately obtain the gray scale density function of an image, and in practice, a histogram of the digital image may be used instead. Histogram enhancement is mainly histogram equalization, and the probability of the gray level histogram of an image after uniform quantization is usually higher in a low-value gray level interval, so that details in a dark area in the image are often not clear. In order to make the image become clear, the gray scale interval of the image can be enlarged, the gray scale with smaller gray scale frequency is enlarged, the histogram of the image is used to replace the distribution density function of the gray scale, and the expression of the function g of the image after histogram equalization is as follows:
where T (r) is a transformation function, r is an integration variable,as a cumulative distribution function of r.
And 2.3, carrying out noise suppression treatment on the enhanced image according to a median filtering method to obtain a noise suppressed image.
Through the median filtering processing, the noise in the image can be obviously suppressed, so that the accuracy of subsequent identification is greatly improved.
And 2.4, carrying out noise elimination processing on the noise suppression image according to a wavelet denoising method to obtain a noise elimination image.
The observation formula assuming the obtained signal is as follows:
yi＝xi+ni，
where ni is white gaussian noise with zero mean, xi is the desired signal, and yi is the observed value.
The filtered noise ni can be considered as how to recover x from the observed value y.
Assuming that the transformation matrix of the discrete wavelet transformation is W, performing wavelet transformation on the above equation yields:
Y＝X+N，
here, Y ═ W [ yi ], X ═ W [ xi ], and N ═ W [ ni ]. For W, there is an inverse transform matrix M, satisfying WM 1.
The wavelet algorithm is utilized to eliminate the noise of the image, so that the noise in the original image can be restrained, the edge details in the image can be reserved, and the subsequent extraction of the image characteristic parameters is facilitated.
Preferably, step 3 specifically comprises:
and 3.1, correcting the preprocessed original image according to the maximum transverse element and the minimum transverse element of the preprocessed original image to obtain a corrected image.
The maximum and minimum number of traverses of the fiber in the entire image, i.e., the noise-canceled image, are first determined and are denoted by max (m1) and min (m1), respectively. And then comparing the size of the columns of the elements with the maximum and minimum transverse rows, if the number of the columns of the pixels with the maximum transverse row is larger than that of the pixels with the minimum transverse row, indicating that the fiber inclines leftwards, and if the number of the columns of the pixels with the maximum transverse row is smaller than that of the pixels with the minimum transverse row, indicating that the fiber inclines rightwards, so that the inclination direction of the fiber can be judged. And finally, determining the rotation direction of the fiber according to the inclined direction of the fiber. The angle is rotated to the left if tilted to the right, and to the right if tilted to the left. angle represents the inclination angle of the fiber in the image, and the size of the angle is atan (l2.l1) × 180/pi, wherein l1 represents the horizontal distance between the leftmost end point and the rightmost end point of the fiber, l2 represents the vertical distance between the lowermost end point and the uppermost end point of the fiber, and a represents the radian of rotation.
And 3.2, determining a circular structural element, opening the corrected image according to the circular structural element, and filtering the corrected image after opening according to a wavelet algorithm to obtain a filtered bubble impurity image.
Firstly, selecting a circular structural element with a smaller diameter, preferably a circular structural element with a diameter of 3 pixels according to the pixel range of the fiber image, then utilizing the structural element to open the fiber image, obviously reducing bubbles and impurities in the fiber image after the opening operation, further utilizing a wavelet algorithm to filter the image after the opening operation, and basically filtering the bubbles and the impurities in the fiber image after the wavelet filtering.
And 3.3, carrying out contour extraction on the filtered bubble impurity image according to a neurodynamic method to obtain a contour extraction image.
The method of contour extraction can be understood as hollowing out the interior points of the target image. If a point in the original image is black and its 8 neighboring points are all black, then the point is an interior point, and the point is removed. It should be noted that although the processed image is a binary image, it is actually a 256-level gray scale image, but only two colors of 0 and 255 are used. Contour tracing is a method for obtaining the contour of an area by tracing point by point according to the connectivity of image boundary points. Different areas on the image can be distinguished by using a contour tracking method, and a basis is provided for further analysis of the image. After obtaining the binary image of the image, firstly, angle correction is carried out on the image to obtain the binary fiber image basically in a vertical or horizontal state. Since the outline of the fiber image is mainly extracted, it is not necessary to extract all the edge information in the image.
Under the above-mentioned premise, it is necessary to first determine whether the fiber is in a horizontal state or a vertical state in the image. If the scanning is in the horizontal state, the image is scanned from top to bottom, from left to right, and the scanning only needs to scan white pixel points in the image, because the outline of the fiber image is completely concentrated at the white pixel points in the image, the scanning is performed from the top left corner of the image from top to bottom, from left to right, all scanning values are sequentially judged, if the scanning is the white pixel points, the scanning of the column is stopped, the horizontal and vertical coordinates of the white pixel points are recorded, then the scanning of the next column is started, and the scanning is performed until the rightmost column of the image is scanned. After the upper half area is scanned, scanning from top to bottom, from left to right from the left lower corner of the image is performed, and the coordinates of each white pixel point are recorded in turn until the rightmost column of the image is scanned. After the scanning is finished, the coordinates of a series of obtained white pixel points are described in another blank image, and the obtained new image is the fiber contour extraction image. Similarly, if the fiber contour extraction image is in a vertical state, scanning is performed from top to bottom, from left to right to obtain coordinates of white pixel points in a series of images, and then the coordinates are redrawn in another blank image according to the obtained coordinates, so that the obtained new image is the contour extraction image of the fiber.
And 3.4, repairing the contour extraction according to a cubic spline interpolation method to obtain a contour repairing image.
Since the contour of the image extracted primarily has discontinuity, namely, broken edge, seamless splicing of the broken edge is required. When repairing the broken edge of the outer contour of the fiber, usually, a possible connection point is found at the end point of the broken part, the end point of the optimal connection can be found by judging the angle, and then the two end points are connected according to the trend of two lines at the two end points.
Two-dimensional matrices are defined for storing the abscissa and ordinate of each end point of the edge line segment in the image. After the whole image is scanned line by line from left to right, short lines with the length less than 10 pixels are removed, and the horizontal and vertical coordinate distribution of the end points of the remaining line segments is stored in the two-dimensional matrixes defined above.
An initial break distance is set, preferably to a distance of 20 pixels, and then starting from the end point of the first line segment, all end points around this end point which are smaller than 20 pixels are searched. Let C be the end point of a line segment, search all possible end points within 20 pixels around C from C, find a satisfactory end point A, regard end point A as the spare end point that end point C may be connected to. And then two points D and B are selected at the positions where the line segment distributed at the endpoint A and the endpoint C is close to the two points. Then the angle values of DCA and BAC are calculated, a threshold value is set, if the difference value of the two angles is less than the threshold value, the two points can be considered to be connected, and then the two points are connected by an interpolation method. According to the steps, the end points which are in accordance with the requirements in the image are sequentially judged and connected, and thus the contour repairing image can be obtained.
And 3.5, carrying out edge detection on the contour patching image according to a Sobel operator method to obtain a boundary image.
The Sobel operator is easy to realize in space, has a certain smooth denoising effect, and can accurately determine the edge direction of the image. The Sobel operator has a local averaging effect, so that the anti-noise performance is better, and the anti-noise performance is increased along with the increase of the used neighborhood. The process of detecting the image edge by using the Sobel operator method is as follows:
a. and filtering and denoising the image by a Gaussian filter.
b. And designing a threshold by using the characteristic that the differential operator highlights the gray level change, and extracting edge points.
c. And connecting and thinning the extracted edge points to obtain the boundary of the image, namely the boundary image.
And 3.6, thinning the boundary image according to a Zhang thinning algorithm to obtain a thinned image.
The morphological method and the Zhang thinning algorithm are combined, and the respective advantages are integrated, so that the image thinning effect is more ideal.
And 3.7, removing burrs in the thinned image according to a set threshold value to obtain the processed image.
After the outer contour of the image is repaired, burrs in the fiber image need to be removed. The fiber edge of the binary image of the fiber image after edge detection has a certain width, and the thinning algorithm eliminates the point with the pixel value of 1 in the fiber image layer by layer, so certain burrs are inevitably generated after the thinning algorithm, and the burrs are long burrs. In addition, noise points and abrupt changes of partial pixels in the image will cause thinned isolated burrs, i.e., short burrs.
For a long spine in an image, one end is the broken end point and the other end is the bifurcation point of the fiber image. Firstly finding out the long thorns from the image, and then removing the long thorns according to the following operations: acquiring a pixel value of a disconnection end point; marking the point, performing tracking search along the broken line, setting a search length threshold, judging that the broken line is a long thorn when an intersection point appears, and directly removing the broken line.
For short spines in the image, both ends of the short spines are end points, so that only one threshold value needs to be set, and short lines between the two ends can be judged as the short spines as long as the threshold value is met, and the short spines can be directly removed.
After the removing operation of the burrs in the fiber image, the outline and the boundary of the fiber image become very clear, and the extraction operation of the characteristic parameters can be more conveniently carried out.
As shown in fig. 2, a is an original image, B is a grayscale image, C is an enhanced image, D is a noise-removed image, E is a corrected image, and F is a processed image.
Preferably, the characteristic parameters include an average twist level, a maximum twist level, a variance of diameters, a maximum-to-minimum diameter ratio, a fullness level, and a variance of fullness level.
Because each characteristic parameter is not in the same order of magnitude and the fluctuation range of each characteristic value is greatly changed, in order to ensure the accuracy of correlation analysis of all characteristic values and the stability of neural network training and operation, firstly, the normalization processing is carried out on each acquired characteristic parameter, and the characteristic parameter values of all samples are ensured to be between [0,1 ]. The formula for parameter normalization is as follows:
wherein X is the characteristic parameter value after normalization processing, X is the characteristic parameter value before normalization processing,is the average value, x, of a certain characteristic parameter to be normalizedmaxIs the maximum value, x, in the characteristic parameterminIs the minimum value in the characteristic parameter.
After the characteristic parameter values of the extracted cotton and linen fibers are subjected to normalization processing, the correlation between the characteristic parameter values and the fiber types needs to be specifically analyzed, the correlation coefficient of each characteristic parameter value is calculated, the characteristic value with good correlation is selected as the basis for identifying the cotton and linen fiber types, and the system accuracy is improved.
Using X to represent a characteristic value vector of the cotton and hemp fibers, n to represent the number of samples, Y to represent the fiber types, sigma X, sigma Y to represent standard deviation, ux，uyExpressing the mean square value, calculating the correlation coefficient R between each characteristic value and the cotton and linen fiber type2Comprises the following steps:
wherein,
and respectively calculating the correlation coefficient between the characteristic parameter of each group of samples after normalization processing and the cotton and linen fiber type, and then averaging to obtain the correlation coefficient between each characteristic value of the cotton and linen fiber sample and the cotton and linen fiber type.
And selecting the average twist degree, the maximum twist degree, the diameter variance, the maximum-minimum diameter ratio, the fullness degree and the fullness degree variance with higher correlation as characteristic parameters for automatically identifying the cotton and linen fiber types through calculation.
Preferably, the step 5 is implemented as follows: and identifying the fiber type of the cotton and linen fibers according to the characteristic parameters by utilizing a BP neural network, and determining the fiber type of the cotton and linen fibers.
The input layer of the BP neural network is used for receiving the foreign data, and the output layer is used for outputting the processing result of the foreign data. The number of neurons in the input layer is set to 6, which corresponds to the 6 characteristic parameters, and the number of neurons in the output layer is set to 1, which corresponds to the type of the cotton and hemp fiber. After the characteristic parameters are normalized, the value range of the parameters of the input samples is consistent with the value range of the S-shaped function, and the values are all between [0 and 1], so the S-shaped function is selected as the transfer function of the hidden layer. If the last layer of the BP neural network is also an S-type function, the output of the whole network is limited to a smaller range. If it is a purely linear function, the input to the whole network can take any value. A purely linear function is chosen as the transfer function of the output layer.
The transfer function of the hidden layer is a logarithmic sigmoid functionThe transfer function of the output layer is a purely linear function f2(x)＝x。
Setting the expected error to be 1 x 10-20. Because the variable quantity of the weight generated by each cycle in the neural network is determined by the learning rate, the whole system is possibly unstable due to the overlarge learning rate, the training time of the network is prolonged and the convergence is slow due to the overlarge learning rate, but a small network error can be ensured. Therefore, on the premise of ensuring the overall stability of the system, a relatively small learning rate is usually selected, and the value range of the learning rate is usually between 0.01 and 0.8. Whether the selected learning rate is proper or not can be judged through the descending speed of the network global error value E after each training. If the descending speed of E is fast, the learning rate is basically selected to be proper, but if the E has oscillation phenomenon, the selected learning rate is selected to be large. Through a large number of repeated experiments and experiences, the initial learning rate is set to be 0.4, and then the learning rate is continuously modified according to the training time and the change size of the global error, so that the network is more stable.
Inputting the characteristic parameters obtained through experiments into a BP neural network according to the set network parameters, training the BP neural network, selecting 50 prepared actual cotton/ramie samples respectively, and testing the BP neural network respectively, wherein the total number of the groups is 10.
And (4) carrying out detection result statistics on actual cotton/ramie fibers by using the trained neural network. The statistics may include the total time consumed for each test and the total number of fibers tested per run, including the number of cotton fibers and ramie fibers. If the ramie fiber is detected, the number of transverse sections and the number of cracks of the ramie fiber can be automatically counted. From the correct recognition rate of the statistical result, the method for automatically detecting the cotton/ramie fiber type has higher practical application efficiency.
As shown in fig. 3, an automatic detecting system for a type of cotton and linen fiber provided by an embodiment of the present invention includes a receiving unit, a preprocessing unit, a processing unit, an extracting unit, and an identifying and counting unit.
The receiving unit is used for receiving the original image of the cotton and linen fibers acquired by the image acquisition device.
And the preprocessing unit is used for preprocessing the original image.
And the processing unit is used for carrying out image processing on the preprocessed original image to obtain a processed image.
The extraction unit is used for extracting the characteristic parameters of the cotton and linen fibers in the processed image.
The identification unit is used for identifying the fiber type of the cotton and linen fibers according to the characteristic parameters.
Preferably, the preprocessing unit is specifically configured to:
and carrying out graying processing on the original image according to a weighted average value method to obtain a grayscale image.
And carrying out image enhancement processing on the gray level image according to a histogram enhancement method to obtain an enhanced image.
And carrying out noise suppression processing on the enhanced image according to a median filtering method to obtain a noise suppressed image.
And carrying out noise elimination processing on the noise suppression image according to a wavelet denoising method to obtain a noise elimination image.
Preferably, the processing unit is specifically configured to:
and correcting the preprocessed original image according to the maximum transverse element and the minimum transverse element of the preprocessed original image to obtain a corrected image.
Determining a circular structural element, performing opening operation on the corrected image according to the circular structural element, and filtering the corrected image after the opening operation according to a wavelet algorithm to obtain a filtered bubble impurity image.
And carrying out contour extraction on the filtered bubble impurity image according to a neurodynamic method to obtain a contour extraction image.
And connecting fracture points by adopting a cubic spline interpolation method according to the image after the contour extraction, and repairing the image after the contour extraction.
And carrying out edge detection on the contour patching image according to a Sobel operator method to obtain a boundary image.
And thinning the boundary image according to a Zhang thinning algorithm to obtain a thinned image.
And removing burrs in the thinned image according to a set threshold value to obtain the processed image.
Preferably, the characteristic parameters include an average twist level, a maximum twist level, a variance of diameters, a maximum-to-minimum diameter ratio, a fullness level, and a variance of fullness level.
Preferably, the processing unit is specifically configured to: and identifying the fiber type of the cotton and linen fibers according to the characteristic parameters by utilizing a BP neural network, and determining the fiber type of the cotton and linen fibers.
The reader should understand that in the description of this specification, reference to the description of the terms "one embodiment," "some embodiments," "an example," "a specific example" or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above are not necessarily intended to refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, various embodiments or examples and features of different embodiments or examples described in this specification can be combined and combined by one skilled in the art without contradiction.
Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention, and that variations, modifications, substitutions and alterations can be made to the above embodiments by those of ordinary skill in the art within the scope of the present invention.
Claims (10)
1. An automatic cotton and linen fiber type detection method is characterized by comprising the following steps:
step 1, receiving an original image of the cotton and linen fibers acquired by an image acquisition device;
step 2, carrying out image preprocessing on the original image;
step 3, carrying out image processing on the preprocessed original image to obtain a processed image;
step 4, extracting characteristic parameters of the cotton and linen fibers in the processed image;
and 5, identifying the fiber type of the cotton and linen fibers according to the characteristic parameters.
2. The method for automatically detecting the type of the cotton and linen fiber according to claim 1, wherein the step 2 specifically comprises:
step 2.1, carrying out graying processing on the original image according to a weighted average value method to obtain a grayscale image;
step 2.2, carrying out image enhancement processing on the gray level image according to a histogram enhancement method to obtain an enhanced image;
step 2.3, carrying out noise suppression processing on the enhanced image according to a median filtering method to obtain a noise suppressed image;
and 2.4, carrying out noise elimination processing on the noise suppression image according to a wavelet denoising method to obtain a noise elimination image.
3. The method for automatically detecting the type of the cotton and linen fiber according to claim 1, wherein the step 3 specifically comprises:
step 3.1, correcting the preprocessed original image according to the maximum transverse element and the minimum transverse element of the preprocessed original image to obtain a corrected image;
step 3.2, determining a circular structural element, performing opening operation on the corrected image according to the circular structural element, and filtering the corrected image after the opening operation according to a wavelet algorithm to obtain a filtered bubble impurity image;
3.3, carrying out contour extraction on the image with the filtered bubble impurities according to a neurodynamic method to obtain a contour extraction image;
step 3.4, the contour extraction is repaired according to a cubic spline interpolation method to obtain a contour repairing image;
step 3.5, carrying out edge detection on the contour patching image according to a Sobel operator method to obtain a boundary image;
step 3.6, thinning the boundary image according to a Zhang thinning algorithm to obtain a thinned image;
and 3.7, removing burrs in the thinned image according to a set threshold value to obtain the processed image.
4. The method according to any one of claims 1 to 3, wherein the characteristic parameters include average twist, maximum twist, variance of diameter, maximum-to-minimum diameter ratio, fullness and variance of fullness.
5. The automatic cotton and linen fiber type detection method according to claim 4, wherein the step 5 is realized by: and identifying the fiber type of the cotton and linen fibers according to the characteristic parameters by utilizing a BP neural network, and determining the fiber type of the cotton and linen fibers.
6. An automatic cotton and linen fiber type detection system is characterized by comprising a receiving unit, a preprocessing unit, a processing unit, an extracting unit and an identifying unit;
the receiving unit is used for receiving the original image of the cotton and linen fibers acquired by the image acquisition device;
the preprocessing unit is used for preprocessing the original image;
the processing unit is used for carrying out image processing on the preprocessed original image to obtain a processed image;
the extraction unit is used for extracting the characteristic parameters of the cotton and linen fibers in the processed image;
the identification unit is used for identifying the fiber type of the cotton and linen fibers according to the characteristic parameters.
7. The automatic cotton and linen fiber type detection system according to claim 6, wherein the preprocessing unit is specifically configured to:
carrying out graying processing on the original image according to a weighted average value method to obtain a grayscale image;
carrying out image enhancement processing on the gray level image according to a histogram enhancement method to obtain an enhanced image;
carrying out noise suppression processing on the enhanced image according to a median filtering method to obtain a noise suppression image;
and carrying out noise elimination processing on the noise suppression image according to a wavelet denoising method to obtain a noise elimination image.
8. The system of claim 6, wherein the processing unit is specifically configured to:
correcting the preprocessed original image according to the maximum transverse element and the minimum transverse element of the preprocessed original image to obtain a corrected image;
determining a circular structural element, performing opening operation on the corrected image according to the circular structural element, and filtering the corrected image after the opening operation according to a wavelet algorithm to obtain a filtered bubble impurity image;
carrying out contour extraction on the filtered bubble impurity image according to a neurodynamic method to obtain a contour extraction image;
repairing the contour extraction according to a cubic spline interpolation method to obtain a contour repairing image;
carrying out edge detection on the contour patching image according to a Sobel operator method to obtain a boundary image;
thinning the boundary image according to a Zhang thinning algorithm to obtain a thinned image;
and removing burrs in the thinned image according to a set threshold value to obtain the processed image.
9. The system of any one of claims 6 to 8, wherein the characteristic parameters include average twist, maximum twist, variance of diameter, maximum-to-minimum diameter ratio, fullness, and variance of fullness.
10. The system of claim 9, wherein the processing unit is specifically configured to: and identifying the fiber type of the cotton and linen fibers according to the characteristic parameters by utilizing a BP neural network, and determining the fiber type of the cotton and linen fibers.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711078382.XA CN108038838A (en) | 2017-11-06 | 2017-11-06 | A kind of cotton fibriia species automatic testing method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711078382.XA CN108038838A (en) | 2017-11-06 | 2017-11-06 | A kind of cotton fibriia species automatic testing method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108038838A true CN108038838A (en) | 2018-05-15 |
Family
ID=62093732
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711078382.XA Pending CN108038838A (en) | 2017-11-06 | 2017-11-06 | A kind of cotton fibriia species automatic testing method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108038838A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109492544A (en) * | 2018-10-19 | 2019-03-19 | 兰波(苏州)智能科技有限公司 | A method of classified by enhancing optical microscopy to animal origin |
| CN109543532A (en) * | 2018-10-19 | 2019-03-29 | 兰波(苏州)智能科技有限公司 | A method of the Fibre sorting based on classification standard and human behavior |
| CN109872287A (en) * | 2019-01-31 | 2019-06-11 | 上海华虹宏力半导体制造有限公司 | The modification method of SEM image line drawing |
| CN110006898A (en) * | 2019-03-28 | 2019-07-12 | 上海工程技术大学 | The blended ratio measuring method of polyester fiber based on circularity algorithm |
| CN110047074A (en) * | 2019-05-17 | 2019-07-23 | 广东工业大学 | The fiber of textile mixes content detection, reverse engineering analysis method and equipment |
| CN111008965A (en) * | 2019-11-28 | 2020-04-14 | 创新奇智(南京)科技有限公司 | System and method for identifying morphological characteristics of cotton-flax blended fibers |
| CN111665244A (en) * | 2019-03-05 | 2020-09-15 | 温力力 | Method based on textile fiber identification and component detection system |
| CN111665243A (en) * | 2019-03-05 | 2020-09-15 | 温力力 | Textile fiber identification and component detection system |
| CN113255790A (en) * | 2021-05-31 | 2021-08-13 | 创新奇智(北京)科技有限公司 | Textile fiber component identification method, electronic device and storage medium |
| CN118154617A (en) * | 2024-05-13 | 2024-06-07 | 菏泽市纺织纤维检验所 | Fiber impurity detection method based on machine vision |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101482927A (en) * | 2009-02-06 | 2009-07-15 | 中国农业大学 | Foreign fiber fuzzy classification system and method based on automatic vision detection |
| CN106373119A (en) * | 2016-09-05 | 2017-02-01 | 广东工业大学 | Fiber detection method and system |
-
2017
- 2017-11-06 CN CN201711078382.XA patent/CN108038838A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101482927A (en) * | 2009-02-06 | 2009-07-15 | 中国农业大学 | Foreign fiber fuzzy classification system and method based on automatic vision detection |
| CN106373119A (en) * | 2016-09-05 | 2017-02-01 | 广东工业大学 | Fiber detection method and system |
Non-Patent Citations (1)
| Title |
|---|
| 赵训明: "《棉与苎麻纤维的计算机图像识别方法的研究》", 《中国优秀硕士学位论文全文数据库工程科技Ⅰ辑》 * |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109492544B (en) * | 2018-10-19 | 2023-01-03 | 兰波(苏州)智能科技有限公司 | Method for classifying animal fibers through enhanced optical microscope |
| CN109543532A (en) * | 2018-10-19 | 2019-03-29 | 兰波(苏州)智能科技有限公司 | A method of the Fibre sorting based on classification standard and human behavior |
| CN109492544A (en) * | 2018-10-19 | 2019-03-19 | 兰波(苏州)智能科技有限公司 | A method of classified by enhancing optical microscopy to animal origin |
| CN109543532B (en) * | 2018-10-19 | 2023-07-28 | 兰波(苏州)智能科技有限公司 | Fiber classification method based on classification standard and human behavior |
| CN109872287A (en) * | 2019-01-31 | 2019-06-11 | 上海华虹宏力半导体制造有限公司 | The modification method of SEM image line drawing |
| CN109872287B (en) * | 2019-01-31 | 2021-01-22 | 上海华虹宏力半导体制造有限公司 | Method for correcting SEM image contour line graph |
| CN111665244B (en) * | 2019-03-05 | 2024-03-19 | 温力力 | Method based on textile fiber identification and component detection system |
| CN111665243B (en) * | 2019-03-05 | 2024-03-19 | 温力力 | Textile fiber identification and component detection system |
| CN111665244A (en) * | 2019-03-05 | 2020-09-15 | 温力力 | Method based on textile fiber identification and component detection system |
| CN111665243A (en) * | 2019-03-05 | 2020-09-15 | 温力力 | Textile fiber identification and component detection system |
| CN110006898A (en) * | 2019-03-28 | 2019-07-12 | 上海工程技术大学 | The blended ratio measuring method of polyester fiber based on circularity algorithm |
| CN110047074A (en) * | 2019-05-17 | 2019-07-23 | 广东工业大学 | The fiber of textile mixes content detection, reverse engineering analysis method and equipment |
| CN111008965A (en) * | 2019-11-28 | 2020-04-14 | 创新奇智(南京)科技有限公司 | System and method for identifying morphological characteristics of cotton-flax blended fibers |
| CN113255790A (en) * | 2021-05-31 | 2021-08-13 | 创新奇智(北京)科技有限公司 | Textile fiber component identification method, electronic device and storage medium |
| CN118154617A (en) * | 2024-05-13 | 2024-06-07 | 菏泽市纺织纤维检验所 | Fiber impurity detection method based on machine vision |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108038838A (en) | A kind of cotton fibriia species automatic testing method and system | |
| CN114937055B (en) | Image self-adaptive segmentation method and system based on artificial intelligence | |
| CN108961217B (en) | Surface defect detection method based on regular training | |
| CN110148130B (en) | Method and device for detecting part defects | |
| CN114419025A (en) | Fiberboard quality evaluation method based on image processing | |
| CN114549522A (en) | Textile quality detection method based on target detection | |
| CN108181316B (en) | Bamboo strip defect detection method based on machine vision | |
| CN111667475B (en) | Machine vision-based Chinese date grading detection method | |
| CN109540925B (en) | Complex ceramic tile surface defect detection method based on difference method and local variance measurement operator | |
| CN113706490B (en) | Wafer defect detection method | |
| CN110749598A (en) | Silkworm cocoon surface defect detection method integrating color, shape and texture characteristics | |
| WO2022267270A1 (en) | Crack characteristic representation method and system based on multi-fractal spectrum | |
| CN117689655B (en) | Metal button surface defect detection method based on computer vision | |
| CN116883408B (en) | Integrating instrument shell defect detection method based on artificial intelligence | |
| CN110807763A (en) | Method and system for detecting ceramic tile surface bulge | |
| CN110648312A (en) | Method for identifying wool and cashmere fibers based on scale morphological characteristic analysis | |
| CN114881965A (en) | Wood board joint detection method based on artificial intelligence and image processing | |
| CN112712058A (en) | Character recognition and extraction method | |
| CN109544513A (en) | A kind of steel pipe end surface defect extraction knowledge method for distinguishing | |
| CN114332079A (en) | Plastic lunch box crack detection method, device and medium based on image processing | |
| CN113673396B (en) | Spore germination rate calculation method, device and storage medium | |
| CN114332081A (en) | Textile surface abnormity determination method based on image processing | |
| CN107239761B (en) | Fruit tree branch pulling effect evaluation method based on skeleton angular point detection | |
| CN117664950A (en) | Innovative drug quality analysis system based on Raman detection technology | |
| CN113256608A (en) | Workpiece defect detection method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |