A method and system for detecting the life of a load cloth for testing the performance of a washing machine

Abstract

The application discloses a kind of load cloth life detection method and system for washing machine performance test, it is related to washing machine performance test field, including, industrial color camera, LED white light bar light source, sample plate, load cloth and computer;The LED white light bar light source is equipped with two, respectively symmetrically arranged in the upper two sides of sample plate, for providing uniform illumination condition;The industrial color camera is vertically arranged above sample plate, lens axis is perpendicular to the center of sample plate, for collecting the digital image of load cloth;The industrial color camera is connected with computer by data line.The application acquires the digital image of load cloth by industrial color camera and symmetrically arranged LED white light bar light source, after computer extracts wrinkle feature and plain weave feature, it is determined using logistic regression classifier to establish life classification model, and the image acquisition quality is ensured by specific size light source and optimized white balance parameter.

Description

Technical Field

[0001] This invention relates to the field of washing machine performance testing, and in particular to a method and system for testing the lifespan of a load cloth used in washing machine performance testing. Background Technology

[0002] In the field of washing machine performance testing, load cloth is used as a standard test medium, and its service life directly affects the accuracy and repeatability of test results. It is required to use mixed load cloths with different wear cycles during testing to simulate actual use conditions. In the existing technology, the wear degree is evaluated by measuring the changes in the physical parameters of the load cloth through contact sensors. Although this method can reflect some wear characteristics, it relies on special equipment and has low detection efficiency, making it difficult to meet the needs of large-scale testing.

[0003] Existing technologies still have significant limitations. Physical performance-based testing methods cannot fully characterize the structural degradation features of load-bearing fabrics. In particular, the correlation between the micro-morphology of the fabric surface and mechanical wear has not been effectively quantified. Manual testing is easily affected by subjective factors and cannot achieve automated batch processing, resulting in low testing efficiency. Existing technologies have not fully utilized the combination of digital image processing and machine learning to achieve automated and accurate quantification of the micro-morphological features of load-bearing fabrics. Among them, wrinkle features are extracted through gray-level co-occurrence matrix and color statistical features, and plain weave features are based on local binary pattern analysis of B / G channel ratio images. A set of intelligent detection systems for load-bearing fabric life has been established, which fundamentally solves the technical problems of chaotic manual recording of load-bearing fabric life and lack of objective evaluation methods in national standard testing. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a method and system for detecting the lifespan of a load-bearing cloth used in washing machine performance testing, which solves the problem that the lifespan of the load-bearing cloth cannot be quantitatively evaluated during washing machine performance testing.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a life testing system for a load cloth used for testing the performance of a washing machine, comprising an industrial color camera, an LED white light strip light source, a sample plate, a load cloth, and a computer;

[0008] Two LED white light strip light sources are provided, symmetrically arranged on both sides above the sample plate to provide uniform illumination.

[0009] The industrial color camera is vertically positioned above the sample plate, with its lens axis perpendicular to the center of the sample plate, and is used to acquire digital images of the load cloth.

[0010] The industrial color camera is connected with a computer through a data line, and the collected image data is transmitted to the computer and processed.

[0011] The computer receives the image data, extracts the wrinkle features and plain weave features of the load cloth, establishes a load cloth life classification model using a logistic regression classifier, and determines the life grade of the load cloth according to the load cloth life classification model.

[0012] As a preferred embodiment of the load cloth life detection system for testing the performance of the washing machine, the light-emitting size of the LED white light strip light source is 360 mm x 30 mm, and the light-emitting area covers the entire working area of the sample plate.

[0013] The industrial color camera adjusts the white balance parameters before collecting the image, wherein the white balance coefficients of the R, G and B channels are 2.0625, 1.0000 and 1.6367, respectively.

[0014] The collected image of the load cloth is folded along the length and width directions to form a sample with a size of one-fourth of the original size.

[0015] During the image collection process of the load cloth, the logo area printed on the surface is avoided.

[0016] As a preferred embodiment of the load cloth life detection system for testing the performance of the washing machine, under the half-load condition of the washing machine, the cotton-60℃ program is run, and every five times, the load cloth is taken out and hung to dry, 10 napkins are taken for digital image collection, the load cloth is flat and has no logo interference, and the long side of the load cloth is parallel to the long side of the LED white light strip light source.

[0017] As a preferred embodiment of the load cloth life detection system for testing the performance of the washing machine, the logistic regression classifier adopts L2 regularization and sets the maximum number of iterations.

[0018] The life classification model divides the load cloth into five life grades corresponding to different running time intervals.

[0019] As a preferred embodiment of the load cloth life detection system for testing the performance of the washing machine, the life grade division is as follows: the first life grade is the load cloth running 5, 10, 15 and 20 times, the second life grade is the load cloth running 25, 30, 35 and 40 times, the third life grade is the load cloth running 45, 50, 55 and 60 times, the fourth life grade is the load cloth running 65, 70, 75 and 80 times, and the fifth life grade is the load cloth running 85, 90, 95 and 100 times.

[0020] In a second aspect, the present application provides a method for detecting the service life of a load cloth for testing the performance of a washing machine, comprising: converting image data into a gray image for adaptive histogram equalization, Gaussian filtering, gradient amplitude calculation and binarization processing, and extracting texture features and color features to obtain wrinkle features;

[0021] A ratio image of the B channel and the G channel is calculated, and after cropping the sub-image, texture features are extracted using a Sobel operator and a local binary pattern to obtain plain weave features.

[0022] As a preferred scheme of the method for detecting the service life of a load cloth for testing the performance of a washing machine, the wrinkle features and the plain weave features are analyzed using a logistic regression classifier to obtain a service life grade determination result of the load cloth.

[0023] As a preferred scheme of the method for detecting the service life of a load cloth for testing the performance of a washing machine, after the load cloth is run for different times according to a standard test procedure, it is placed flat on a sample plate, an LED white light bar light source is turned on, the white balance parameters of an industrial color camera are adjusted, digital images of the load cloth are collected, image data is transmitted to a computer, and wrinkle features and plain weave features are extracted through an image processing algorithm.

[0024] As a preferred scheme of the method for detecting the service life of a load cloth for testing the performance of a washing machine, 80% of the image data in each service life grade is extracted as a training set, and 20% is extracted as a test set, a logistic regression classifier is trained using the training set, a load cloth service life grading model is established, the accuracy of the model is verified using the test set, and the parameters of the logistic regression classifier are optimized.

[0025] As a preferred scheme of the method for detecting the service life of a load cloth for testing the performance of a washing machine, according to the confusion matrix result of the load cloth service life grading model, the feature extraction algorithm is adjusted to improve the prediction accuracy, and the feature extraction strategy for different load cloth types is optimized.

[0026] In a third aspect, the present application provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and wherein the computer program is executed by the processor to implement any step of the method for detecting the service life of a load cloth for testing the performance of a washing machine according to the first aspect of the present application.

[0027] In a fourth aspect, the present application provides a computer readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement any step of the method for detecting the service life of a load cloth for testing the performance of a washing machine according to the first aspect of the present application.

[0028] The present application has the beneficial effects that: through the industrial color camera and the symmetrically arranged LED white light strip light source, the digital image of the load cloth is collected, after the computer extracts the wrinkle features and the plain weave features, the life grading model is established by using the logic regression classifier to determine, the image collection quality is ensured through the specific size of the light source and the optimized white balance parameters, the features are extracted by using the image processing technology, and the load cloth is divided into five life grades, the automatic, non-contact and accurate detection of the load cloth life is realized, and the problems that the traditional method cannot quantize the microscopic morphology change and is low in efficiency are solved. BRIEF DESCRIPTION OF DRAWINGS

[0029] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiment description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0030] Fig. 1 The flow chart of the life detection system of the load cloth for the performance test of the washing machine;

[0031] Fig. 2 The flow chart of the life detection method of the load cloth for the performance test of the washing machine;

[0032] Fig. 3 The structure schematic diagram of the life detection system of the load cloth for the performance test of the washing machine;

[0033] Fig. 4 The wrinkle image feature extraction flow chart of the load cloth for the life detection method of the load cloth for the performance test of the washing machine;

[0034] Fig. 5 The plain weave image feature extraction flow chart of the load cloth for the life detection method of the load cloth for the performance test of the washing machine;

[0035] Fig. 6 The confusion matrix schematic diagram of the load life grading model.

[0036] In the figure, 1 is an industrial color camera, 2 is an LED white light strip light source, 3 is a sample plate, 4 is a load cloth, and 5 is a computer. DETAILED DESCRIPTION

[0037] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail in combination with the drawings of the specification.

[0038] In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application. However, it will be apparent to one skilled in the art that the present application can be practiced without the specific details set forth in this description, that the present application can be practiced with other than the described implementations, and that the present application can be practiced with other than the described implementations, and that the present application can be practiced in other environments.

[0039] Secondly, the "one embodiment" or "embodiment" referred to herein means a specific feature, structure, or characteristic that can be included in at least one implementation of the present application. The "in one embodiment" appearing in different places in the specification does not mean the same embodiment, nor is it an embodiment that is separate or alternative to other embodiments.

[0040] Embodiment 1, reference Figs. 1-6 As a first embodiment of the present application, the embodiment provides a load cloth life detection system for testing the performance of a washing machine, comprising the following steps:

[0041] The light-emitting size of the LED white light strip light source 2 is 360 mm × 30 mm, and the light-emitting area covers the entire working area of the sample plate 3.

[0042] Furthermore, the LED white light strip light source 2 adopts a light-emitting size of 360 mm × 30 mm, which ensures that the light-emitting area completely covers the working area of the sample plate 3, providing uniform and stable lighting conditions for digital image acquisition of the load cloth 4. The size design enables two symmetrically arranged LED white light strip light sources 2 to form a shadow-free lighting environment along the axis direction of the industrial color camera 1, while avoiding image feature extraction errors caused by uneven lighting. The long strip structure of the light source is arranged in parallel with the long side of the load cloth 4, further optimizing the imaging effect of the surface texture features.

[0043] Before collecting the image, the industrial color camera 1 adjusts the white balance parameters, wherein the white balance coefficients of the R, G, and B channels are 2.0625, 1.0000, and 1.6367, respectively.

[0044] Furthermore, the industrial color camera 1 sets the R channel white balance coefficient to 2.0625, the G channel to 1.0000, and the B channel to 1.6367 before image acquisition. By precisely adjusting the gain ratio of the three channels to compensate for the spectral characteristic differences of the LED white light strip light source 2, the true restoration of the surface color features of the load cloth 4 is ensured, and the color reproducibility under the 360 mm × 30 mm LED white light strip light source 2 irradiation condition is optimized. This enables the industrial color camera 1 to accurately capture the original color information of the load cloth 4, providing a colorimetric accurate image data basis for subsequent wrinkle feature and plain weave feature extraction. The white balance adjustment is performed immediately after the LED white light strip light source 2 is turned on, eliminating environmental light interference and maintaining the consistency of the imaging conditions of all life grade load cloths 4.

[0045] The loaded cloth 4 is folded in the length and width directions respectively to form a sample to be measured with a size of one fourth of the original size.

[0046] Further, the loaded cloth 4 is folded once in the length and width directions respectively before image acquisition to form a flat sample to be measured with an area of one fourth of the original size, so that the size of the measured area of the loaded cloth 4 is adapted to the 1280x1024 pixel imaging range of the industrial color camera 1, and at the same time, the additional wrinkle interference caused by multiple folding is eliminated. The long side of the folded loaded cloth 4 is placed parallel to the long side of the LED white light strip light source 2, ensuring uniform light distribution in the 360mmx30mm illumination area, providing a standardized sample morphology for subsequent wrinkle feature and plain weave feature extraction. The pre-processing method improves the feature distinguishability of the image collected by the industrial color camera 1 while maintaining the integrity of the wear features of the loaded cloth 4.

[0047] During the image acquisition process of the loaded cloth 4, the logo area printed on the surface is avoided.

[0048] Further, during the image acquisition process of the loaded cloth 4, the viewfinder range of the industrial color camera 1 actively avoids the logo area printed on the surface of the loaded cloth 4, ensuring that the collected image only contains the pure cotton fabric area of the loaded cloth 4. By adjusting the placement position of the loaded cloth 4 on the sample plate 3, the 1280x1024 pixel imaging area of the industrial color camera 1 completely covers the logo-free fabric surface while the long side of the loaded cloth 4 is parallel to the long side of the LED white light strip light source 2, effectively avoiding the interference of the logo pattern on the extraction of wrinkle features and plain weave features.

[0049] Under the half-load condition of the washing machine, run the cotton-60°C program, take out the loaded cloth every five times and hang it to dry, take 10 napkins for digital image acquisition, the loaded cloth 4 is flat and has no logo interference, and the long side of the loaded cloth 4 is parallel to the long side of the LED white light strip light source 2.

[0050] Further, under the condition of the washing machine half load, the cotton-60℃ standard test procedure is performed, the load cloth 4 is taken out after every 5 washing cycles are completed and is hung to dry in a flat state, 10 napkin samples are selected from each taken-out load cloth 4 to perform digital image acquisition, during the acquisition process, it is ensured that the load cloth 4 is completely flattened and the surface is not disturbed by the logo pattern, and meanwhile, the long edge direction of the load cloth 4 is strictly kept parallel to the long edge direction of the LED white light strip light source 2, under the uniform illumination of the LED white light strip light source 2, the industrial color camera 1 acquires a 1280*1024 pixel digital image according to the preset white balance parameters (R channel 2.0625, G channel 1.0000, B channel 1.6367), so as to ensure that the image data of the load cloth 4 at different life stages is comparable, and to establish a consistent experimental basis for subsequent feature extraction and life grading.

[0051] The logistic regression classifier adopts L2 regularization and sets the maximum number of iterations.

[0052] Further, the logistic regression classifier adopts the L2 regularization method in the training process, adds the sum of squares of weight parameters in the loss function to control the model complexity and prevent overfitting, and sets the maximum number of iterations to 1000 (example value), so as to immediately terminate the optimization process when the training iteration reaches the number, ensure that the model training is completed under limited computing resources, and the L2 regularization applies constraints to the weight parameters, so that the logistic regression classifier maintains the generalization ability when fitting the wrinkle feature and the plain weave feature.

[0053] The life grading model divides the load cloth into five life grades, corresponding to different running number intervals.

[0054] Further, the grading standard is based on the wrinkle feature and the plain weave feature extracted from the digital image of the load cloth 4 collected by the industrial color camera 1, the classification boundary is established through the training of the logistic regression classifier, each grade corresponds to a different mechanical wear feature distribution, and the output result of the life grading model is directly related to the actual use state of the load cloth in the washing machine performance test.

[0055] The life grades are divided into the first life grade, the load cloth running for 5, 10, 15 and 20 times, the second life grade, the load cloth running for 25, 30, 35 and 40 times, the third life grade, the load cloth running for 45, 50, 55 and 60 times, the fourth life grade, the load cloth running for 65, 70, 75 and 80 times, and the fifth life grade, the load cloth running for 85, 90, 95 and 100 times.

[0056] Embodiment 2, refer to Figs. 1-6 As a second embodiment of the present application, the embodiment also provides a life detection method for a load cloth for washing machine performance test, comprising:

[0057] S1. Convert the image data into a grayscale image, perform adaptive histogram equalization, Gaussian filtering, gradient magnitude calculation and binarization, and extract texture features and color features to obtain wrinkle features.

[0058] Specifically, the grayscale conversion expression is:

[0059] ;

[0060] in, These are the pixel values ​​of a grayscale image. The red channel pixel values ​​of the original color image. The green channel pixel value of the original color image. This represents the blue channel pixel value of the original color image.

[0061] Specifically, the adaptive histogram equalization expression is as follows:

[0062] ;

[0063] in, For the adaptive histogram equalized image in coordinates Pixel value at that location, For grayscale images in coordinates The original pixel value at that location, These are the two-dimensional coordinates of pixels in the image. This is a local transformation function.

[0064] Specifically, the Gaussian filter expression is:

[0065] ;

[0066] in, The image after Gaussian filtering is in coordinates Pixel value at that location, For Gaussian kernel in offset The weight value at that location, Let be the radius of the Gaussian kernel. For the input image after adaptive histogram equalization, in coordinates Pixel value at that location, coordinates The neighborhood offset.

[0067] Specifically, the expression for the gradient magnitude is:

[0068] ;

[0069] in, Image at pixel coordinates gradient magnitude at that point The gradient is in the horizontal direction. This represents the gradient in the vertical direction.

[0070] Specifically, the binary expression is:

[0071] ;

[0072] in, For binarized images in coordinates Pixel value at that location, The gradient magnitude in coordinates The value at that location,

[0073] It should be noted that the digital image of the load cloth 4 acquired by the industrial color camera 1 is first converted into a grayscale image. The image contrast is enhanced by adaptive histogram equalization, and then a smooth image is obtained by Gaussian filtering. After gradient magnitude calculation, the smooth image is binarized. Simultaneously, the entropy value of the pixel value of the grayscale image, the grayscale co-occurrence matrix features (including contrast, correlation, energy and uniformity), and the total area of ​​the connected regions of the binarized image are extracted as texture features. At the same time, the color mean, standard deviation, skewness and kurtosis of the binarized image are calculated as color features. The above processing flow completely preserves the wrinkle morphology features of the surface of the load cloth 4 caused by mechanical wear. A quantitative index characterizing the wear degree of the load cloth 4 is constructed by multi-dimensional feature fusion.

[0074] The ratio of the B channel to the G channel is calculated, and after cropping the sub-image, the texture features are extracted using the Sobel operator and local binary mode to obtain the plain weave texture features.

[0075] Specifically, the ratio graph expression is as follows:

[0076] ;

[0077] in, for, Original image Channel in coordinates Pixel value at that location, Original image Channel in coordinates The pixel value at that location.

[0078] Specifically, the sub-image cropping expression is:

[0079] ;

[0080] in, For the first Zhang's scissors image, These are the coordinates of the center of the ratio graph.

[0081] Specifically, the Sobel edge detection expression is,

[0082] ;

[0083] wherein, is the edge intensity map of the i-th sub-image, is the horizontal Sobel operator, is the vertical Sobel operator, is the sub-image index.

[0084] Specifically, the binarization processing expression is,

[0085] ;

[0086] wherein, is the binarized edge map of the i-th sub-image, is the edge intensity value of the i-th sub-image. Further, the G channel and the B channel of the original image are extracted to generate a ratio image, the flat weave features are strengthened and the wrinkle interference is weakened by dividing the pixel value of the B channel by the pixel value of the G channel, four sub-images with a size of 640x512 pixels are cropped based on the center of the ratio image, the Sobel operator edge detection is performed on each sub-image to obtain a directional filtering result, the local binary pattern texture features of each sub-image are calculated after the binarization processing, and the average value of the local binary pattern features of the four sub-images is taken as the flat weave feature.

[0087] The embodiment also provides a life detection method of the load cloth for performance testing of the washing machine, comprising:

[0088] S2, the wrinkle feature and the flat weave feature are analyzed by using the logistic regression classifier to obtain a life grade determination result of the load cloth 4.

[0089] S2, the wrinkle feature and the flat weave feature are analyzed by using the logistic regression classifier to obtain a life grade determination result of the load cloth 4.

[0090] Further, the single-channel gray image obtained after the gray conversion of the original color image of the load cloth 4 collected by the industrial color camera 1 is generated by linear combination calculation of the original RGB three channels according to the weight coefficient (red channel 0.299, green channel 0.587, and blue channel 0.114), the value range is 0 (pure black) to 255 (pure white), in the load cloth life detection process, the pixel value of the gray image is used as the basic image data for subsequent adaptive histogram equalization and Gaussian filter processing, and directly participates in the extraction of the texture features (entropy, gray level co-occurrence matrix features) and the color features (mean value, standard deviation, skewness, and kurtosis), and is the core input for quantifying the wrinkle and wear degree of the surface of the load cloth 4.​​

[0091] S3, after the load cloth 4 is run for different times according to the standard test procedure, the load cloth 4 is placed flat on the sample plate 3, the LED white light bar light source 2 is turned on, the white balance parameters of the industrial color camera 1 are adjusted, the digital image of the load cloth 4 is collected, the image data is transmitted to the computer 5, and the wrinkle feature and the plain weave feature are extracted through the image processing algorithm.

[0092] Further, after the load cloth 4 is run for a specified number of times according to the standard test procedure, the load cloth 4 is placed in a fully expanded state on the sample plate 3, under the symmetrical illumination of the LED white light bar light source 2, the industrial color camera 1 collects a 1280x1024 pixel digital image according to the preset white balance parameters (R channel 2.0625, G channel 1.0000, B channel 1.6367), the industrial color camera 1 transmits the image to the computer 5 through the data line, and the computer 5 performs the following processing flow to convert the original image to a grayscale image and perform adaptive histogram equalization, extract the wrinkle feature through Gaussian filtering and gradient amplitude calculation, and based on the B / G ratio image, crop four 640x512 sub-images, and extract the plain weave feature through Sobel operator and local binary pattern processing.

[0093] It should be noted that the load cloth 4 must be fully expanded and laid flat on the sample plate 3 after completing the standard test procedure to ensure that there is no folding or twisting to eliminate the interference of additional wrinkles on the detection result, the two light sources of the LED white light bar light source 2 need to symmetrically illuminate the sample plate 3 at a 45-degree angle to form a uniform shadowless lighting environment, the long side of the light source must be strictly parallel to the long side of the load cloth 4, the industrial color camera 1 needs to be accurately white balanced before collection, and the R, G, and B channel coefficients are fixed at 2.0625, 1.0000, and 1.6367, respectively, to compensate for the spectral characteristics of the LED light source, the logo area on the surface of the load cloth 4 needs to be avoided during the collection process to ensure that the image only contains the pure cotton fabric part, the 1280x1024 pixel image received by the computer 5 needs to be geometrically corrected to eliminate the influence of lens distortion before subsequent feature extraction, when extracting the wrinkle feature, the neighborhood size of adaptive histogram equalization should be set to 15x15 pixels, and the sigma value of Gaussian filtering should be 1.5 to balance noise suppression and feature preservation, in the extraction of the plain weave feature, the B / G ratio image needs to be gamma corrected to enhance the texture information in the low-contrast area, the sub-image cropping needs to ensure that the four areas cover the main wear parts of the load cloth 4, and all feature parameters need to be standardized by Z-score to eliminate the influence of dimension difference on the classifier.

[0094] S4, extract 80% of the image data in each life level as the training set and 20% as the test set, train the logistic regression classifier with the training set, establish the load cloth life level classification model, verify the accuracy of the model using the test set, and optimize the parameters of the logistic regression classifier.

[0095] Further, the digital image data of the load cloth 4 collected in each life level (first to fifth life level) is randomly extracted in a proportion of 80% as the training set and the remaining 20% as the test set. The wrinkle features and plain weave features contained in the training set are input into the logistic regression classifier for training. The classifier adopts L2 regularization to constrain the weight parameters, sets the maximum number of iterations to 1000 times, optimizes the model parameters by minimizing the cross-entropy loss function, establishes the load cloth life level classification model, and uses the test set to evaluate the performance of the model, calculate the classification accuracy and verify the generalization ability of the model. According to the test results, adjust the regularization coefficient and other hyperparameters of the logistic regression classifier. The optimized load cloth life level classification model can determine the life level of unknown load cloth 4 samples, and the output result corresponds to five life level classifications of 5-100 runs, ensuring that the model not only learns the association rules between features and life levels, but also avoids overfitting the training data.

[0096] S5, according to the confusion matrix result of the load cloth life level classification model, adjust the feature extraction algorithm to improve the prediction accuracy, and optimize the feature extraction strategy for different types of load cloth 4.

[0097] Further, analyze the confusion matrix result output by the load cloth life level classification model, re-evaluate the extraction process of the wrinkle features and plain weave features for the misclassified samples, optimize the displacement parameters and direction value combinations of the gray level co-occurrence matrix for the frequently misclassified life levels, and improve the calculation radius and neighborhood point number of the local binary pattern feature. In view of the material differences of different types of load cloth 4 (napkin, bed sheet, shirt, handkerchief), while keeping the core algorithms (Sobel operator, adaptive histogram equalization, etc.) unchanged, the ratio weight of B channel to G channel, the parameters of Gaussian filter and the value of binary threshold are adjusted respectively, so that the feature extraction process adapts to the surface characteristics of various types of load cloth 4. The optimized feature extraction strategy is verified by retraining the logistic regression classifier to ensure that the load cloth life level classification model maintains high prediction accuracy for all types of load cloth 4.

[0098] The embodiment also provides a computer device suitable for the load cloth life detection method for washing machine performance test, which comprises a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions to realize the load cloth life detection method for washing machine performance test proposed in the above embodiment.

[0099] The computer device can be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner. The wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball or touchpad arranged on the shell of the computer device, or an external keyboard, touchpad or mouse, etc.

[0100] In summary, the present application collects digital images of the loaded cloth by an industrial color camera and a symmetrically arranged LED white light strip light source, and after the computer extracts wrinkle features and plain weave features, a life grading model is established by using a logistic regression classifier to determine the life grading model, the image acquisition quality is ensured by a specific size of the light source and optimized white balance parameters, the features are extracted by using image processing technology, and the loaded cloth is divided into five life grades, so that the automatic and non-contact precise detection of the life of the loaded cloth is realized.

[0101] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application but not limit the present application. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present application, and all of them should be covered in the scope of the claims of the present application.

Claims

1. A lifespan testing system for a load-bearing cloth used in washing machine performance testing, characterized in that: Includes an industrial color camera (1), an LED white light strip light source (2), a sample plate (3), a load cloth (4), and a computer (5); The LED white light strip light source (2) has a light-emitting size of 360 mm × 30 mm, and the light-emitting area covers the entire working area of ​​the sample plate (3); Before acquiring images, the industrial color camera (1) adjusts the white balance parameters, wherein the white balance coefficients of the R, G, and B channels are 2.0625, 1.0000, and 1.6367, respectively. The load cloth (4) is folded in half along the length and width directions to form a test sample with a size one-quarter of the original size; During the image acquisition process of the load cloth (4), the logo area printed on the surface is avoided; Two LED white light strip light sources (2) are provided, which are symmetrically arranged on both sides above the sample plate (3) to provide uniform lighting conditions. The industrial color camera (1) is vertically arranged above the sample plate (3), with the lens axis perpendicular to the center of the sample plate (3), and is used to acquire digital images of the load cloth (4); The industrial color camera (1) is connected to the computer (5) via a data cable to transmit the acquired image data to the computer (5) for processing; The computer (5) receives image data, extracts the fold features and plain weave features of the load fabric (4) through image processing algorithms, establishes a load fabric life grading model using a logistic regression classifier, and determines the life grade of the load fabric (4) according to the load fabric life grading model.

2. The lifespan testing system for load-bearing cloth used in washing machine performance testing as described in claim 1, characterized in that: Under half-load conditions in the washing machine, the cotton -60℃ program is run. The load cloth is taken out and hung to dry every five times. Ten load cloths (4) are taken for digital image acquisition. The load cloth (4) is flat and has no logo interference. The long side of the load cloth (4) is parallel to the long side of the LED white light strip light source (2).

3. The lifespan testing system for load-bearing cloth used in washing machine performance testing as described in claim 2, characterized in that: The logistic regression classifier uses L2 regularization and sets a maximum number of iterations; The lifespan grading model divides the load distribution into five lifespan levels, corresponding to different ranges of operating times.

4. The life testing system for load cloth used in washing machine performance testing as described in claim 3, characterized in that: The lifespan classification is as follows: first lifespan level, load cloths that have been run for 5, 10, 15, or 20 cycles; second lifespan level, load cloths that have been run for 25, 30, 35, or 40 cycles; third lifespan level, load cloths that have been run for 45, 50, 55, or 60 cycles; fourth lifespan level, load cloths that have been run for 65, 70, 75, or 80 cycles; and fifth lifespan level, load cloths that have been run for 85, 90, 95, or 100 cycles.

5. A method for detecting the lifespan of a load-bearing cloth used for testing the performance of a washing machine, based on the lifespan detection system for the load-bearing cloth used for testing the performance of a washing machine as described in any one of claims 1 to 4, characterized in that: include, The image data is converted into a grayscale image and subjected to adaptive histogram equalization, Gaussian filtering, gradient magnitude calculation and binarization. Texture and color features are then extracted to obtain wrinkle features. The ratio of the B channel to the G channel is calculated, and after cropping the sub-image, the texture features are extracted using the Sobel operator and local binary mode to obtain the plain weave texture features.

6. The method for testing the lifespan of a load-bearing cloth used in washing machine performance testing as described in claim 5, characterized in that: This includes using a logistic regression classifier to analyze the wrinkle features and plain weave features to obtain the lifespan rating of the load fabric (4).

7. The method for testing the lifespan of a load-bearing cloth used in washing machine performance testing as described in claim 6, characterized in that: The process includes running the load cloth (4) according to the standard test procedure a different number of times, placing it flat on the sample plate (3), turning on the LED white light strip light source (2), adjusting the white balance parameters of the industrial color camera (1), acquiring digital images of the load cloth (4), transmitting the image data to the computer (5), and extracting wrinkle features and plain weave features through image processing algorithms.

8. The method for testing the lifespan of a load-bearing cloth used in washing machine performance testing as described in claim 7, characterized in that: This involves extracting 80% of the image data from each lifespan level as a training set and 20% as a test set. The training set is used to train a logistic regression classifier to build a load cell lifespan grading model. The test set is used to verify the accuracy of the model and optimize the parameters of the logistic regression classifier.

9. The method for testing the lifespan of a load-bearing cloth used in washing machine performance testing as described in claim 8, characterized in that: This includes adjusting the feature extraction algorithm based on the confusion matrix results of the load fabric lifetime grading model to improve prediction accuracy, and optimizing the feature extraction strategy for different load fabric (4) types.

Images