Method and device for detecting oil state of range hood and intelligent range hood

Abstract

The application relates to an oil stain state detection method and device of an extractor hood and an intelligent extractor hood. The oil stain state detection method of the extractor hood extracts an oil stain coverage rate and a gray scale distribution change index from a target gray scale image; and according to a set scale and direction, a multi-scale texture roughness of the target gray scale image is extracted based on a gray scale co-occurrence matrix; the scale and direction are set based on internal oil stain adhesion characteristics of the extractor hood; according to oil stain sample data covering different materials and / or structures, a regression optimization algorithm is combined to determine a first fusion weight corresponding to the oil stain coverage rate, a second fusion weight corresponding to the gray scale distribution change index and a third fusion weight corresponding to the texture roughness; the oil stain coverage rate, the gray scale distribution change index and the texture roughness are weighted and fused according to the first fusion weight, the second fusion weight and the third fusion weight, so as to obtain an oil stain index of the extractor hood.

Description

Technical Field

[0001] This application relates to the field of range hoods, and in particular to methods, devices, and intelligent range hoods for detecting the oil stain status of range hoods. Background Technology

[0002] Range hoods are essential kitchen appliances for daily cooking. As users pay more attention to the user experience, they are also demanding higher levels of intelligence from range hoods. Inside the range hood, especially the fan impeller and grease filter, condensation easily forms when hot and humid fumes are exhausted and then cool down. This condensation, when mixed with grease, exacerbates grease buildup. Furthermore, external air humidity also affects the condensation rate, especially in humid weather, which increases grease buildup. Therefore, regular cleaning of the range hood is necessary.

[0003] Current range hood cleaning reminders are often based on the motor's operating time reaching a corresponding cleaning cycle. This method, which sets cleaning reminders according to motor operating time, fails to adapt to the differences in grease buildup caused by varying cooking habits in different regions. In areas where steaming and boiling are the primary cooking methods, even with minimal grease buildup after prolonged use at low operating speeds, regular cleaning is still necessary. However, in areas where stir-frying is the primary cooking method, the range hood, under high load or high back pressure, will develop odors or become contaminated with grease even after short periods of use. The inability to clean it promptly reduces the range hood's smoke extraction capacity. In response, some manufacturers have shortened the overall maintenance cycle, which increases maintenance costs, and the cleaning reminders remain inaccurate and untimely.

[0004] There is currently no effective solution to the problem of untimely cleaning reminders for range hoods in related technologies. Summary of the Invention

[0005] This embodiment provides a method, device, and intelligent range hood for detecting the oil stain status of a range hood, in order to solve the problem of untimely cleaning reminders for range hoods in related technologies.

[0006] Firstly, this embodiment provides a method for detecting the oil stain status of a range hood, including:

[0007] With the range hood turned off, acquire a target grayscale image of the inside of the fan housing of the range hood;

[0008] The oil stain coverage rate and gray-level distribution change index are extracted from the target gray-level image; and, based on the gray-level co-occurrence matrix, the multi-scale texture roughness of the target gray-level image is extracted according to the set scale and direction; the scale and direction are set based on the internal oil stain adhesion characteristics of the range hood.

[0009] Based on oil stain sample data covering different materials and / or structures, and combined with regression optimization algorithms, the first fusion weight corresponding to the oil stain coverage rate, the second fusion weight corresponding to the gray scale distribution change index, and the third fusion weight corresponding to the texture roughness are determined.

[0010] The oil stain coverage, the grayscale distribution change index, and the texture roughness are weighted and fused according to the first fusion weight, the second fusion weight, and the third fusion weight to obtain the oil stain index of the range hood.

[0011] This embodiment can accurately detect the state of oil stains and provide cleaning reminders for range hoods that are adapted to different usage habits and intensities, thereby improving the timeliness of cleaning reminders.

[0012] In some embodiments, extracting oil stain coverage and grayscale distribution change index from the target grayscale image includes:

[0013] Based on the baseline clean state grayscale image, the oil stain feature region of the target grayscale image is determined, and the oil stain coverage rate is extracted based on the oil stain feature region;

[0014] Calculate the first gray-level histogram of the baseline clean state gray-level image and the second gray-level histogram of the target gray-level image respectively. Based on the first gray-level histogram and the second gray-level histogram, determine the gray-level distribution change index of the target gray-level image.

[0015] This embodiment can quantify the extent of oil spillage inside the range hood and the overall grayscale feature changes caused by the oil spill based on grayscale feature extraction.

[0016] In some embodiments, based on a baseline clean state grayscale image, oil stain feature regions of the target grayscale image are determined, and the oil stain coverage is extracted based on the oil stain feature regions, including:

[0017] Calculate the pixel color difference image between the target grayscale image and the baseline clean state grayscale image;

[0018] Based on the inter-class variance, an adaptive segmentation threshold for the pixel color difference image is determined;

[0019] Based on the adaptive segmentation threshold, the oil stain feature region is segmented from the pixel color difference image;

[0020] The oil stain coverage rate is determined by the ratio between the number of pixels in the oil stain feature area and the total number of pixels in the pixel color difference image.

[0021] This embodiment can accurately detect oil pollution coverage, providing a stable data foundation for subsequent determination of the oil pollution index.

[0022] In some embodiments, determining the grayscale distribution change index of the target grayscale image based on the first grayscale histogram and the second grayscale histogram includes:

[0023] Calculate the chi-square distance between the first gray-level histogram and the second gray-level histogram, and the rate of change of the mean gray level between the first gray-level histogram and the second gray-level histogram;

[0024] The gray-scale distribution change index is obtained by weighting and fusing the chi-square distance and the gray-scale mean change rate.

[0025] This embodiment enables accurate assessment of changes in grayscale distribution.

[0026] In some embodiments, multi-scale texture roughness of the target grayscale image is extracted based on the gray-level co-occurrence matrix according to a set scale and orientation, including:

[0027] Based on the physical size characteristics of oil stains on the impeller surface of range hoods collected in history, microscale, mesoscale, and macroscale are set for gray-level co-occurrence matrix calculation; based on the distribution characteristics of oil stains inside range hoods collected in history, multiple directions are set for gray-level co-occurrence matrix calculation.

[0028] For the set microscale, mesoscale, and macroscale, calculate the corresponding gray-level co-occurrence matrix under each set direction;

[0029] All gray-level co-occurrence matrices are fused to obtain multi-scale texture roughness.

[0030] This embodiment can improve the accuracy of texture roughness calculation for different oil stain adhesion stages by using micro, meso and macro scales and combining four directions based on the characteristics of oil stains at different adhesion stages.

[0031] In some of these embodiments, all gray-level co-occurrence matrices are fused to obtain multi-scale texture roughness, including:

[0032] For each of the set scales, the average roughness of the set multiple directions is calculated based on the probability matrix of the gray-level co-occurrence matrix corresponding to each direction.

[0033] The average roughness of each scale is weighted and fused to obtain multi-scale texture roughness.

[0034] This embodiment can achieve weighted fusion of roughness at different scales, thereby improving the accuracy and stability of the final comprehensive texture roughness.

[0035] In some embodiments, based on oil stain sample data covering different materials and / or structures, and combined with a regression optimization algorithm, a first fusion weight corresponding to the oil stain coverage rate, a second fusion weight corresponding to the grayscale distribution change index, and a third fusion weight corresponding to the texture roughness are determined, including:

[0036] Based on historically collected grayscale images of internal samples of range hoods with different materials and / or structures, combined with labeled oil stain level tags, oil stain sample data is constructed; the sample grayscale images include several sets of images, each set of images being grayscale images of the same range hood sample under different oil stain levels, and the materials and / or structures of the range hood samples in different sets of images are different.

[0037] Based on the oil pollution sample data and the ridge regression optimization algorithm, the first fusion weight, the second fusion weight, and the third fusion weight are optimized to obtain the optimized first fusion weight, the second fusion weight, and the third fusion weight.

[0038] This embodiment, by setting weighting coefficients differently, can adapt to the characteristics of different range hood products and improve the accuracy and universality of oil stain level determination.

[0039] In some embodiments, after obtaining the oil pollution index, the method further includes:

[0040] Based on the calibrated mathematical mapping relationship between the oil stain index and air volume compensation, the fan compensation speed corresponding to the oil stain index is determined; the mathematical mapping relationship uses the oil stain index as the independent variable and the fan compensation speed as the dependent variable, and the mathematical mapping relationship includes the range hood preset speed and compensation coefficient.

[0041] The range hood's speed setting for the next startup is updated based on the fan's compensated rotation speed.

[0042] This embodiment can also automatically increase the fan speed according to the value of the oil stain index to compensate for the decrease in aerodynamic performance caused by oil stains. It solves the problem that fixed-cycle maintenance in related technologies cannot adapt to different usage intensities. It achieves the dual effect of accurate oil stain prediction and cleaning maintenance of the range hood, as well as adaptive compensation of performance, so that the cleaning cycle is extended by more than 30% while still having good oil fume suction power.

[0043] Secondly, this embodiment provides a range hood oil stain status detection device, including: an acquisition module, a feature extraction module, a weight determination module, and a weighted fusion module; wherein:

[0044] The acquisition module is used to acquire a target grayscale image inside the fan housing of the range hood when the range hood is turned off.

[0045] The feature extraction module is used to extract the oil stain coverage rate and gray-level distribution change index from the target gray-level image; and, based on the gray-level co-occurrence matrix, extract the multi-scale texture roughness of the target gray-level image according to the set scale and direction; the scale and direction are set based on the internal oil stain adhesion characteristics of the range hood.

[0046] The weight determination module is used to determine, based on oil stain sample data covering different materials and / or structures, and in conjunction with a regression optimization algorithm, the first fusion weight corresponding to the oil stain coverage rate, the second fusion weight corresponding to the grayscale distribution change index, and the third fusion weight corresponding to the texture roughness.

[0047] The weighted fusion module is used to perform weighted fusion of the oil stain coverage, the gray scale distribution change index and the texture roughness according to the first fusion weight, the second fusion weight and the third fusion weight to obtain the oil stain index of the range hood.

[0048] Thirdly, this embodiment provides an intelligent range hood, including a range hood body and the oil stain status detection device for the range hood described in the second aspect.

[0049] Compared with related technologies, this embodiment provides a method, device, and intelligent range hood for detecting the oil stain status of a range hood. The method for detecting the oil stain status of a range hood involves acquiring a target grayscale image of the inside of the range hood's fan housing when the range hood is off; extracting the oil stain coverage rate and grayscale distribution change index from the target grayscale image; and extracting multi-scale texture roughness of the target grayscale image based on a grayscale co-occurrence matrix according to a set scale and direction; the scale and direction are set based on the internal oil stain adhesion characteristics of the range hood; determining a first fusion weight corresponding to the oil stain coverage rate, a second fusion weight corresponding to the grayscale distribution change index, and a third fusion weight corresponding to the texture roughness based on oil stain sample data covering different materials and / or structures, combined with a regression optimization algorithm; and weighting and fusing the oil stain coverage rate, grayscale distribution change index, and texture roughness according to the first, second, and third fusion weights to obtain the oil stain index of the range hood. This method can accurately detect the oil stain status and provide range hood cleaning reminders adapted to different usage habits and intensities, improving the timeliness of range hood cleaning reminders.

[0050] Details of one or more embodiments of this application are set forth in the following drawings and description to make other features, objects and advantages of this application more readily apparent. Attached Figure Description

[0051] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0052] Figure 1 This is an application scenario diagram of an embodiment of this application;

[0053] Figure 2 This is a schematic diagram of a control structure for detecting oil pollution status according to an embodiment of this application;

[0054] Figure 3 This is a flowchart of the oil stain detection method for a range hood according to an embodiment of this application;

[0055] Figure 4 This is a schematic diagram of a signal flow according to an embodiment of this application;

[0056] Figure 5 This is a flowchart illustrating a multi-scale texture roughness extraction method according to an embodiment of this application.

[0057] Figure 6 This is a flowchart of a method for detecting the oil stain status of a range hood according to some embodiments of this application;

[0058] Figure 7 This is a structural block diagram of the oil stain status detection device for a range hood according to an embodiment of this application. Detailed Implementation

[0059] To better understand the purpose, technical solution, and advantages of this application, the application is described and illustrated below in conjunction with the accompanying drawings and embodiments.

[0060] Unless otherwise defined, the technical or scientific terms used in this application shall have the general meaning understood by one of ordinary skill in the art to which this application pertains. Words such as “a,” “an,” “an,” “the,” “the,” and “these” used in this application do not indicate quantitative limitation and may be singular or plural. The terms “comprising,” “including,” “having,” and any variations thereof used in this application are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that comprises a series of steps or modules (units) is not limited to the listed steps or modules (units) but may include steps or modules (units) not listed, or may include other steps or modules (units) inherent to these processes, methods, products, or devices. Words such as “connected,” “linked,” and “coupled” used in this application are not limited to physical or mechanical connections but may include electrical connections, whether direct or indirect. “Multiple” used in this application refers to two or more. “And / or” describes the relationship between related objects, indicating that three relationships may exist; for example, “A and / or B” can represent: A alone, A and B simultaneously, and B alone. Normally, the character " / " indicates that the objects before and after it are in an "or" relationship. The terms "first," "second," "third," etc., used in this application are merely to distinguish similar objects and do not represent a specific order of objects.

[0061] Figure 1 This is an application scenario diagram of this embodiment, such as... Figure 1 As shown, inside the fan housing 10 of the range hood, there are structures such as a volute 11, a motor 12, and an impeller 13. In this embodiment, a camera is installed inside the fan housing 10 of the range hood to capture images of the interior of the fan housing. This can be achieved as follows: Figure 1 A camera and supplementary lighting integrated assembly 14 is installed in areas facing key components such as the volute 11 and impeller 13 to capture images of the structural surfaces of these components. Exemplarily, the camera can be a miniature (cylindrical, 8 mm in diameter and 10 mm in length) complementary metal-oxide-semiconductor (CMOS) camera, with the lens facing the plane of rotation of the volute 11 and impeller 13, covering a field of view of ≥120°. Furthermore, the camera can be equipped with an oleophobic cover, such as a high-temperature resistant window with an oleophobic coating on the outer surface, capable of withstanding temperatures above 80°C. The supplementary lighting can be an LED supplementary lighting, such as a white LED with a color temperature of 6000 Kelvin. The supplementary lighting is integrated near the camera and illuminates for 200 milliseconds per shot.

[0062] Figure 2 This is a schematic diagram of a control structure for detecting oil contamination in this embodiment, as shown below. Figure 2As shown, this control structure can be installed inside the range hood, and specifically includes a main controller 21, a light module 22, a camera 23, a supplementary light 24, a storage module 25, a switch module 26, a fan drive module 27, a communication module 28, and an optional lifting drive module 29. The fan drive module 27 includes sensors for detecting speed, current, and voltage; the optional lifting drive module 29 can drive the smoke hood or air inlet to rise or fall. The main controller 21 sends control signals and communicates with the light module 22, camera 23, supplementary light 24, storage module 25, switch module 26, fan drive module 27, communication module 28, and optional lifting drive module 29. The main controller 21 is responsible for timing control and image processing algorithms, and makes decisions regarding cleaning reminders, fan compensation control, and user interface prompts.

[0063] This embodiment provides a method for detecting the oil stain status of a range hood. Figure 3 This is a flowchart of the oil stain detection method for the range hood in this embodiment, as shown below. Figure 3 As shown, the process includes the following steps:

[0064] Step S301: With the range hood off, acquire a target grayscale image of the inside of the range hood's fan housing. Specifically, this can be done after a preset time period (e.g., 30 seconds) following each shutdown of the range hood, allowing the grease to settle, before activating the camera to capture an RGB image of the inside of the fan housing. First, the raw RGB signal acquired by the camera can be a two-dimensional digital matrix composed of M×N pixels, where each pixel contains intensity values ​​for the R, G, and B channels, represented as:

[0065] ;

[0066] Where (x, y) are the coordinates of the pixel in the image, x ∈ [0, M-1], y ∈ [0, N-1]. R raw G raw And B raw These represent the original brightness values ​​of the corresponding pixels in the red, green, and blue color channels, ranging from 0 to 255. These values ​​are directly derived from the photoelectric conversion output of the CMOS sensor. The acquired raw RGB signal is an unprocessed physical signal, containing various components such as oil stains, metallic backgrounds, ambient lighting, lens dust, and circuit noise. The raw RGB image is represented as follows:

[0067] ;

[0068] R(x,y), G(x,y), and B(x,y) represent the intensity of pixel (x,y) in the R, G, and B channels, respectively, output by the CMOS sensor; t represents the acquisition timestamp, obtained based on the real-time clock (RTC).

[0069] Then, the original RGB image is converted to a grayscale image, which can be represented as:

[0070] ;

[0071] In the formula, R, G, and B represent the RGB channel intensities.

[0072] The grayscale image is then denoised. Specifically, an adaptive Wiener filter can be used for pre-processing the grayscale image to achieve denoising, resulting in:

[0073] ;

[0074] Where μ is the local mean, which can be calculated for pixels within a window of a local preset size (e.g., 7×7); For local variance; v 2 The noise variance is obtained through dark field calibration.

[0075] In this way, the target grayscale image can be obtained.

[0076] Step S302: Extract the oil stain coverage rate and gray-level distribution change index from the target gray-level image; and extract the multi-scale texture roughness of the target gray-level image based on the gray-level co-occurrence matrix according to the set scale and direction; the scale and direction are set based on the internal oil stain adhesion characteristics of the range hood.

[0077] The oil stain coverage rate specifically represents the ratio of pixels belonging to the oil stain area in the target grayscale image to the total number of pixels in the target grayscale image. As the most intuitive physical quantity, the oil stain coverage rate reflects the extent of oil stain spread within the monitored area. The oil stain area can be obtained by comparing the target grayscale image with a pre-acquired baseline clean state image and performing binarization processing on this basis. The baseline clean state image is a grayscale image collected when the inside of the range hood is in a clean state. The grayscale distribution change index can be used to characterize the statistical index of the grayscale change of the target grayscale image compared to the baseline clean state image. Among them, when oil stains adhere to the metal or coated surfaces such as the volute and impeller inside the range hood, it will cause a systematic change in the surface optical properties of these structures. Specifically, for clean surfaces, after metal galvanizing or sandblasting, spraying, etc., a Lambertian reflector is often formed, resulting in a higher overall grayscale value and a relatively more concentrated distribution. The grayscale value follows a Gaussian distribution.

[0078] ;

[0079] in, This represents the average grayscale value of the cleaned surface. This represents the variance of the grayscale values ​​of the cleaned surface.

[0080] For surfaces with oil stains, the grease absorbs specific wavelengths of light, resulting in a decrease in overall grayscale. Simultaneously, the uneven distribution of oil stains leads to an increase in grayscale variance, meaning the grayscale distribution becomes more dispersed. Based on this, the dimensionless grayscale distribution change index can be used to quantify the changes in overall grayscale characteristics caused by oil stain adhesion.

[0081] Furthermore, multiple scales for calculating the gray-level co-occurrence matrix can be set based on the characteristics of oil adhesion inside the range hood, such as the common physical size characteristics of oil on the impeller surface. Moreover, the directions used for calculating the gray-level co-occurrence matrix can be set based on the texture distribution of oil flowing in different directions after gradual accumulation; that is, based on the distribution characteristics of oil inside the range hood in different directions, the aforementioned multiple directions can be set. Therefore, by using multi-scale texture roughness, the different stages of oil adhesion can be accurately captured, improving the accuracy of determining the degree of oil adhesion.

[0082] Step S303: Based on the oil stain sample data covering different materials and / or structures, and combined with the regression optimization algorithm, determine the first fusion weight corresponding to the oil stain coverage rate, the second fusion weight corresponding to the gray scale distribution change index, and the third fusion weight corresponding to the texture roughness.

[0083] Regression optimization algorithms can also be used to determine the fusion weights corresponding to the above indicators based on oil stain sample data covering different materials and / or structures. Specifically, the first fusion weight corresponding to the oil stain coverage rate, the second fusion weight corresponding to the grayscale distribution change index, and the third fusion weight corresponding to the texture roughness are determined. Specifically, this embodiment considers the differences in surface characteristics, clean surface grayscale features, and oil stain adhesion characteristics of different range hood materials and structures. The influence of the oil stain coverage rate, grayscale distribution change index, and texture roughness varies on different materials and structures. Therefore, corresponding fusion weights are pre-set for the corresponding materials and structures based on oil stain sample data covering different materials and / or structures. For example, the differences in commonly used materials for range hoods are shown in Table 1:

[0084] Table 1

[0085]

[0086] The differences in the common internal structures of range hoods are shown in Table 2:

[0087] Table 2

[0088]

[0089] Combining Tables 1 and 2, considering the differences in structure and materials of range hoods, the degree to which the aforementioned oil stain coverage rate, grayscale distribution variation index, and texture roughness affect the calculation of the oil stain index varies for different range hoods. Therefore, corresponding oil stain sample data can be constructed for range hood samples with different materials and structures.

[0090] Specifically, several range hood samples of different materials and / or structures can be selected. For each range hood sample, multiple grayscale images with different levels of grease are taken, and the corresponding grease levels are manually labeled. Then, a regression optimization algorithm is used to optimize the optimal values ​​of the first, second, and third fusion weights for the corresponding material and / or structure. Taking range hoods of different materials as examples, the optimal values ​​of the optimized first, second, and third fusion weights are shown in Table 3.

[0091] Table 3

[0092]

[0093] As shown in Table 3, since the Teflon coating surface is relatively smooth, oil stains mainly cover the surface. For range hoods made of this type of material, the first fusion weight corresponding to the oil stain coverage rate is the largest (0.45). Since the brushed stainless steel material has directional texture, the third fusion weight corresponding to the texture roughness is the largest (0.40). Since the sandblasted aluminum alloy material has a rough surface, both the third fusion weight and the first fusion weight are relatively large.

[0094] In other words, the oil stain coverage rate reflects the proportion of oil stain area, and this indicator is applicable to materials with smooth surfaces where oil stains mainly cover the surface; the gray scale distribution change index reflects the overall brightness change, and is applicable to materials with uniform surfaces and strong reflectivity; the texture roughness reflects the changes in the surface microstructure, and is applicable to materials with rough surfaces and textures.

[0095] This embodiment differentiates the above-mentioned fusion weights based on the differences in the structure and / or materials of range hoods, which can adapt to the characteristics of different range hood products and improve the accuracy and universality of the oil stain index calculation.

[0096] Step S304: According to the first fusion weight, the second fusion weight and the third fusion weight, the oil stain coverage, gray scale distribution change index and texture roughness are weighted and fused to obtain the oil stain index of the range hood.

[0097] After determining the first, second, and third fusion weights, the oil pollution index I can be calculated. oil :

[0098] ;

[0099] Where γ1 represents the first fusion weight; γ2 represents the second fusion weight; γ3 represents the third fusion weight; C v Indicates oil pollution coverage rate, Represents the grayscale distribution change index, R C,total This represents texture roughness. For example, based on optimization, the first fusion weight is 0.4, the second fusion weight is 0.3, and the third fusion weight is 0.3.

[0100] Furthermore, after obtaining the aforementioned oil pollution index, the oil pollution index is compared with the index ranges corresponding to multiple preset oil pollution levels to determine the oil pollution level corresponding to the currently calculated oil pollution index. For example, Table 4 below can be used to determine the oil pollution level:

[0101] Table 4

[0102]

[0103] When the oil stain index is ≥0.25 (oil stain level L2 or above), a cleaning reminder is required. A tiered cleaning reminder strategy can be set based on the different oil stain levels. For example, when the oil stain level is L0 to L1, only the oil stain index and level are recorded, and no cleaning reminder is issued; when the oil stain level is L2, a reminder is issued suggesting cleaning soon; when the oil stain level is L3, a reminder is issued suggesting cleaning this week; and when the oil stain level is L4, a reminder is issued suggesting immediate cleaning.

[0104] In addition, the airflow loss can be determined based on the oil stain index, and then the airflow setting of the range hood can be compensated for the next use.

[0105] Therefore, compared to related technologies that rely on motor runtime for regular cleaning reminders, this embodiment transforms traditional time-based maintenance into state-based maintenance. By extracting three types of indicators from grayscale images inside the range hood—representing the degree of oil spread, changes in grayscale features caused by oil adhesion, and texture distribution caused by oil adhesion—oil condition detection is achieved based on user usage data. This results in a quantified oil index mapping oil level, forming graded cleaning reminders. This enables dynamic cleaning reminders based on usage data, reducing odor buildup from condensed fumes, extending maintenance cycles, and mitigating the inaccuracies of traditional cleaning reminders caused by differences or changes in user habits in different regions. Consequently, different range hoods can adaptively trigger cleaning reminders according to changes in usage intensity and habits, ensuring better fume extraction and a healthier kitchen environment with minimal delay, thus improving the range hood's adaptability to various scenarios and enhancing the user's cooking and cleaning experience.

[0106] Therefore, through the above steps S301 to S304, a target grayscale image of the inside of the range hood's fan housing is acquired while the range hood is off. The oil stain coverage rate and grayscale distribution change index are extracted from the target grayscale image. Furthermore, based on a set scale and direction, multi-scale texture roughness of the target grayscale image is extracted using a grayscale co-occurrence matrix. The scale and direction are set based on the internal oil stain adhesion characteristics of the range hood. Based on oil stain sample data covering different materials and / or structures, and combined with a regression optimization algorithm, a first fusion weight corresponding to the oil stain coverage rate, a second fusion weight corresponding to the grayscale distribution change index, and a third fusion weight corresponding to the texture roughness are determined. The oil stain coverage rate, grayscale distribution change index, and texture roughness are weighted and fused according to the first, second, and third fusion weights to obtain the range hood's oil stain index. This allows for accurate detection of oil stain conditions, enabling range hood cleaning reminders adapted to different usage habits and intensities, thus improving the timeliness of range hood cleaning reminders.

[0107] In one embodiment, extracting the oil stain coverage rate and gray distribution change index from the target grayscale image may specifically include:

[0108] Based on the baseline clean state grayscale image, the oil stain feature region of the target grayscale image is determined, and the oil stain coverage rate is extracted based on the oil stain feature region. The first grayscale histogram of the baseline clean state grayscale image and the second grayscale histogram of the target grayscale image are calculated respectively. Based on the first grayscale histogram and the second grayscale histogram, the grayscale distribution change index of the target grayscale image is determined.

[0109] The target grayscale image can be compared with the baseline clean state grayscale image, and the oil stain feature area can be determined based on the comparison result. The ratio of the number of pixels in the oil stain feature area to the total number of pixels is determined as the oil stain coverage rate, which reflects the degree of spread of oil stains inside the range hood.

[0110] Furthermore, based on the gray-level histogram, the gray-level distribution change index of the target gray-level image relative to the baseline clean state gray-level image can be determined to reflect the overall gray-level feature changes caused by oil contamination. Specifically, the gray-level histogram of the baseline clean state gray-level image can be extracted as the first gray-level histogram, and the gray-level histogram of the target gray-level image can be extracted as the second gray-level histogram. Based on the differences in shape and brightness between the first and second gray-level histograms, the aforementioned gray-level distribution change index is determined.

[0111] This embodiment can quantify the extent of oil spillage inside the range hood and the overall grayscale feature changes caused by the oil spill based on grayscale feature extraction.

[0112] More specifically, in one embodiment, based on a baseline clean state grayscale image, the oil stain feature regions of the target grayscale image are determined, and the oil stain coverage is extracted based on the oil stain feature regions. This may specifically include:

[0113] Calculate the pixel color difference image between the target grayscale image and the baseline clean state grayscale image; determine the adaptive segmentation threshold of the pixel color difference image based on the inter-class variance; segment the oil stain feature region from the pixel color difference image based on the adaptive segmentation threshold; determine the oil stain coverage rate by the ratio between the number of pixels in the oil stain feature region and the total number of pixels in the pixel color difference image.

[0114] First, the target grayscale image Grayscale image of baseline cleaning state Comparison:

[0115] ;

[0116] in, The calculated pixel color difference image is then used. An adaptive segmentation threshold is set for the pixel color difference image based on the inter-class variance. Specifically, this embodiment considers that differences in different user kitchen environments and the cleanliness of the camera lens itself can affect the overall color baseline of the acquired image, and a fixed segmentation threshold is prone to misjudgment. Therefore, the entire pixel color difference image can be treated as a grayscale image, and a binarization segmentation algorithm (Otsu) is used to traverse all possible candidate thresholds T to determine an optimal threshold that maximizes the inter-class variance between the background (low color difference) and foreground (high color difference) pixel classes separated by this optimal threshold.

[0117] During the calculation, the normalized histogram p(i) of the pixel color difference image is first determined, where i is the color difference level (0 to 255). Color difference levels with a non-zero probability of occurrence in the normalized histogram can be used as candidate thresholds. For each candidate threshold T, the intra-class probability of the corresponding foreground pixels is calculated (i.e., the proportion of foreground pixels to the total number of pixels, i.e., the proportion of pixels with a color difference level ≥ T). ;

[0118] Calculate the intra-class probability of the corresponding background pixels (i.e., the proportion of background pixels to the total number of pixels, i.e., the proportion of pixels with color difference levels from 0 to T): ;

[0119] Calculate the average color difference of the foreground: ;

[0120] Calculate the average color difference of the background: ;

[0121] Calculate the between-class variance: ;

[0122] This will cause inter-class variance The largest candidate threshold T opt This optimal threshold is identified as the adaptive segmentation threshold.

[0123] Then, based on the adaptive segmentation threshold, the pixel color difference image is binarized:

[0124] ;

[0125] in, Coordinates in a pixel color difference image pixel values, The set of pixels with a median value of "1" represents the initially segmented oil stain feature area (foreground).

[0126] Then calculate The number of pixels with a median value of 1, i.e., the number of oily pixels. ,use Represent it. Calculate the total number of pixels in the pixel color difference image. Calculate the oil pollution coverage rate C v for:

[0127] ;

[0128] Oil coverage rate directly corresponds to the proportion of surface area covered by oil. The larger the value, the wider the area of ​​the surface of interest covered by oil. This embodiment can accurately detect oil coverage rate, providing a stable data foundation for subsequent determination of the oil contamination index.

[0129] In another embodiment, determining the grayscale distribution change index of the target grayscale image based on the first grayscale histogram and the second grayscale histogram may specifically include:

[0130] Calculate the chi-square distance between the first and second gray-level histograms, and the rate of change of the mean gray-level between the first and second gray-level histograms; weight and fuse the chi-square distance and the rate of change of the mean gray-level to obtain the gray-level distribution change index.

[0131] First, the gray-level histogram of the baseline clean state gray-level image is calculated as the first gray-level histogram. :

[0132] ;

[0133] Calculate the gray-level histogram of the target gray-level image as the second gray-level histogram. :

[0134] ;

[0135] Where k is the gray value of the corresponding pixel; Let be the Kronecker function, and when the condition is true, then ,otherwise .

[0136] Normalize the first and second gray-level histograms into probability distributions:

[0137] ;

[0138] Next, the chi-square distance between the two probability distributions is calculated to measure the difference between them:

[0139] ;

[0140] The reason for using chi-square distance in calculating the chi-square distance is... Instead of a single probability, this is to avoid the division-by-zero problem and provide stability for low-probability values. The chi-square distance (numerical range [0,∞)) is a dimensionless parameter that reflects changes in the shape of the grayscale distribution. For clean surfaces, the grayscale distribution is concentrated and highly reflective; for surfaces with oil or grease, the grayscale distribution becomes more dispersed.

[0141] Alternatively, the grayscale mean can be calculated based on the above probability distribution:

[0142] ;

[0143] Then calculate the rate of change of the mean gray level:

[0144] ;

[0145] The grayscale mean change rate (numerical range [0,1]) can reflect the overall brightness change. Oil stains will reduce the surface reflectivity and cause the average grayscale to decrease.

[0146] Then, the chi-square distance and the rate of change of the mean gray level are linearly combined to obtain the gray level distribution change index (with a numerical range of [0,1]):

[0147] ;

[0148] The weighting coefficients α and β can be empirical values, or they can be calibrated by those skilled in the art to other values, such as α=0.6 and β=0.4. This embodiment can achieve accurate assessment of grayscale distribution changes.

[0149] Table 5 shows typical values ​​for chi-square distance, grayscale mean change rate, and grayscale distribution change index for clean surfaces and heavily oiled surfaces, respectively:

[0150] Table 5

[0151]

[0152] Furthermore, in one embodiment, based on a set scale and orientation, multi-scale texture roughness of the target grayscale image is extracted using a grayscale co-occurrence matrix, which may specifically include:

[0153] Based on the physical size characteristics of oil stains on the impeller surface of range hoods collected historically, micro, meso, and macro scales were defined for calculating the gray-level co-occurrence matrix. According to the distribution characteristics of oil stains inside the range hoods collected historically, multiple directions were defined for calculating the gray-level co-occurrence matrix. For each defined micro, meso, and macro scale, the corresponding gray-level co-occurrence matrix was calculated under each defined direction. All gray-level co-occurrence matrices were then fused to obtain multi-scale texture roughness.

[0154] Specifically, this embodiment sets the scale and direction for grayscale co-occurrence matrix calculation based on the actual physical size characteristics of oil stains on the surface of the range hood impeller. In the initial stage of oil adhesion, it adheres to the metal surface in the form of a molecular layer, only tens of micrometers thick, forming an oil film that is difficult to detect with the naked eye. At this time, the texture of the oil stains exhibits micrometer-level inhomogeneity, requiring a microscale (d=1) pixel distance to capture it. As cooking times increase, oil molecules gradually condense into oil droplets, reaching a diameter of approximately 0.1 to 0.5 millimeters. At this point, the spot-like texture clusters formed by the condensation of oil droplets can be captured using a mesoscale (d=4) pixel distance; this is a key characteristic of the transformation of oil stains from a uniform oil film to non-uniform condensation. After long-term accumulation of oil stains, they form directional flow marks under gravity, reaching lengths on the centimeter scale. At this point, a macroscale (d=10) pixel distance can be used to detect these large-scale, directional texture patterns, reflecting the solidification and agglomeration state of the oil stains. Therefore, the scales are set to d=1, d=4, and d=10 respectively.

[0155] In addition, this embodiment selects four angles (90°, 0°, 45°, and 135°). Among them, the 90° direction is the vertical direction, which is consistent with the direction of gravity and can reflect the downward flow of oil stains under the influence of gravity. It can detect the texture distribution caused by the gradual accumulation of oil stains flowing along the direction of gravity. The 0° direction is the horizontal direction, which is perpendicular to the direction of gravity and is used to detect the texture distribution of oil stains along the horizontal direction. The 45° and 135° are diagonal directions and are used to detect anisotropic characteristics and determine whether the oil stains have other typical directions (such as flow marks).

[0156] The distribution characteristics of oil stains on the inner surface of the range hood differ in different directions:

[0157] For clean surfaces, the texture is isotropic, and the roughness values ​​in the four directions mentioned above are not significantly different.

[0158] The initial oil film also exhibits isotropic texture with minimal differences in the four directions;

[0159] For the intermediate-stage oil droplets, slight anisotropy begins to appear;

[0160] For late-stage flow marks, the texture roughness in the gravity direction (90° direction) is significantly greater than in other directions.

[0161] Compared to other fields that typically use conventional gray-level co-occurrence matrices (GLCMs) with only one scale (e.g., d=1) and two directions (e.g., 0° and 90°), or employ a symmetrical approach (e.g., calculating only 0° and 90°), this embodiment is based on the analysis of the oil stain distribution characteristics at different stages on a range hood. It specifically uses the aforementioned four angles for GLCM calculation, covering all possible texture directions to detect anisotropic features. Furthermore, it innovates in scale selection, choosing three physically meaningful scales corresponding to three different physical stages of oil stain adhesion. By combining each scale with each direction, 12 GLCMs can be formed, comprehensively capturing the texture features of the oil stains.

[0162] Specifically, in the initial stage of oil stain adhesion (e.g., after approximately 0 to 50 uses of a range hood), during cooking, oil molecules are adsorbed onto the metal surface in monolayer form, forming a uniform thin film. The texture characteristics change in this stage as follows: at the microscale (d=1), the texture roughness increases from 5-10 during cleaning to 10-15, an increase of approximately 50%; this is because the oil film disrupts the uniform reflection of the metal surface, producing slight differences in grayscale. At the mesoscale (d=4), the texture roughness does not change significantly (increasing from 8-12 to 9-13, an increase of approximately 10%); this is because the oil film is uniform and oil droplets have not yet coalesced. At the macroscale (d=10), the roughness remains almost unchanged (increasing from 3.6 to 3.7, an increase of approximately 5%), because the oil film thickness is uniform and there are no obvious flow marks. Therefore, in the initial stage of oil film adhesion, the roughness increases significantly mainly at the microscale, while the roughness changes little at the mesoscale and macroscale.

[0163] During the middle stage of oil contamination (approximately 50 to 200 uses), oil droplets coalesce, oil molecules continue to accumulate, and surface tension causes the oil film to rupture, forming tiny oil droplets with a diameter of approximately 0.1 to 0.5 millimeters. At this stage, at the microscale, roughness continues to increase (e.g., from 10-15 to 15-20, an increase of approximately 40%), due to more abrupt grayscale changes at the droplet edges. At the mesoscale, roughness begins to increase significantly (from 9-13 to 14-20, an increase of approximately 50%), because the droplet diameter of approximately 0.2 millimeters matches a pixel distance of d=4, allowing the grayscale co-occurrence matrix to capture the grayscale differences between droplets. At the macroscale, roughness increases slightly (e.g., from 3-7 to 5-9, an increase of approximately 30%), because the oil droplets are relatively evenly distributed and have not yet formed large-scale patterns. Therefore, during the middle stage of oil contamination, roughness at the mesoscale begins to increase significantly, narrowing the gap with roughness at the microscale.

[0164] In the late stage of oil contamination (after approximately 200 uses), the flow marks solidify. Oil droplets converge and flow under gravity, forming directional flow marks, while the oil contamination begins to solidify. At the microscale, roughness tends to saturate (e.g., increasing from 15-20 to 18-22, an increase of about 15%) because the surface is almost completely covered by oil, leaving limited room for further increase. At the mesoscale, roughness continues to increase (from 14-20 to 18-25, an increase of about 25%) because the oil droplets further coalesce and increase in size. At the macroscale, roughness increases significantly (from 5-9 to 10-16, an increase of about 80%) due to the formation of flow marks, exhibiting large-scale texture patterns on the centimeter scale. At this point, the roughness difference in the four directions also increases significantly; the roughness ratio in the 0° direction to the 90° direction can reach 1.5 to 2.0, indicating the formation of directional flow marks, with the vertical direction significantly affected by gravity. Therefore, in the late stage of oil contamination, roughness at the macroscale increases significantly, and anisotropy is evident.

[0165] Therefore, roughness at the microscale Roughness at the mesoscale and roughness at the macroscopic scale When combined, they can form a comprehensive judgment logic as shown in Table 6:

[0166] Table 6

[0167]

[0168] Compared to single-scale methods that can only detect oil films at a certain stage and cannot distinguish between different stages of oil films, and where intermediate-stage oil droplets and late-stage flow marks are similar in appearance at the microscale and easily confused, this embodiment distinguishes different development stages of oil pollution by different scales of sensitivity: the microscale focuses on the initial stage, the mesoscale focuses on the intermediate stage, and the macroscale focuses on the late stage. Through anisotropy analysis, the directional flow characteristics of oil pollution can be detected.

[0169] Specifically, after setting the above scales and directions, to reduce the amount of computation, the 256 gray levels can be reduced to L levels (let L=16 or 32, or 256 levels can be retained if the computing power is sufficient):

[0170] ;

[0171] Then, the gray-level co-occurrence matrix (GLCM) is calculated, where for each scale d... i With each direction ,calculate First, create an L×L zero matrix P. For each pixel (x, y) in the image, calculate the positions of neighboring pixels based on their orientation and distance:

[0172] ;

[0173] ;

[0174] ;

[0175] ;

[0176] Then, the co-existence pairs are statistically analyzed: if Within the image range, then:

[0177] ;

[0178] ;

[0179] Then P is normalized into a probability matrix:

[0180] ;

[0181] Where m and n are gray level indices; then, the roughness calculation formula is used to calculate a certain angle in a certain direction d. Roughness:

[0182] ;

[0183] The physical meaning of the roughness formula: ;

[0184] Among molecules Emphasizing the contribution of large grayscale differences, corresponding to the edges of oil stains and abrupt texture changes; denominator Suppress the influence of small grayscale differences (noise); the roughness formula is sensitive to grayscale differences between the second and fourth orders, making it more suitable for oily texture features; i and j represent grayscale level indices.

[0185] Then, the roughness at each scale and in each direction is fused to obtain multi-scale texture roughness.

[0186] This embodiment can improve the accuracy of texture roughness calculation for different oil stain adhesion stages by using micro, meso and macro scales and combining four directions based on the characteristics of oil stains at different adhesion stages.

[0187] In one embodiment, fusing all gray-level co-occurrence matrices to obtain multi-scale texture roughness may include:

[0188] For each set scale, the average roughness of multiple set directions is calculated based on the probability matrix of the gray-level co-occurrence matrix in each corresponding direction; the average roughness of each scale is weighted and fused to obtain the multi-scale texture roughness.

[0189] First, after obtaining the gray-level co-occurrence matrix, the roughness at each scale and in each direction can be calculated based on the roughness formula. Then, for each scale d... i Calculate the average roughness in four directions:

[0190] ;

[0191] Then, weighted fusion is performed on the different scales:

[0192] ;

[0193] In this study, the weight w1 corresponding to the microscale can be 0.3, the weight w2 corresponding to the mesoscale can be 0.5, and the weight w3 corresponding to the macroscale can be 0.2. This yields the multi-scale texture roughness.

[0194] This embodiment can achieve weighted fusion of roughness at different scales, thereby improving the accuracy and stability of the final comprehensive texture roughness.

[0195] Furthermore, in one embodiment, based on oil stain sample data covering different materials and / or structures, and combined with a regression optimization algorithm, a first fusion weight corresponding to the oil stain coverage rate, a second fusion weight corresponding to the grayscale distribution change index, and a third fusion weight corresponding to the texture roughness are determined, which may specifically include:

[0196] Based on historically collected grayscale images of internal samples of range hoods with different materials and / or structures, combined with labeled oil stain level tags, oil stain sample data is constructed. The sample grayscale images include several sets of images, each set being a grayscale image of the same range hood sample under different oil stain levels. The materials and / or structures of the range hood samples in different sets of images are different. Based on the oil stain sample data and the ridge regression optimization algorithm, the first fusion weight, the second fusion weight, and the third fusion weight are optimized to obtain the optimized first fusion weight, the second fusion weight, and the third fusion weight.

[0197] Specifically, 10 range hood samples of different materials and / or structures can be selected. For each range hood sample, 100 images with different levels of grease are taken and processed into grayscale images to obtain the grayscale images of the internal samples of the range hoods. Then, the grease levels (e.g., L0 to L4) of these 100 grayscale images of the internal samples of the range hoods are manually labeled to construct grease sample data. The grease state detection method described above can be applied to the grease sample data. Based on the detected grease index and the manually labeled grease levels, the first, second, and third fusion weights are optimized using a ridge regression optimization algorithm. The specific expressions are as follows:

[0198] ;

[0199] Where z represents the index of the grayscale image sample inside the range hood, Z represents the number of grayscale images sampled inside the range hood, and y z The value of manually labeled oil contamination level normalized to the [0,1] interval can be specifically represented as:

[0200] ;

[0201] Where L z The level is manually labeled (values ​​0, 1, 2, 3, 4), L max =4. Therefore, the converted value is 4. , The oil pollution index, This is the regularization parameter.

[0202] This embodiment, by setting weighting coefficients differently, can adapt to the characteristics of different range hood products and improve the accuracy and universality of oil stain level determination.

[0203] In another embodiment, after obtaining the oil pollution index, the above-mentioned oil pollution status detection method may further include:

[0204] Based on the calibrated mathematical mapping relationship between the oil stain index and air volume compensation, the fan compensation speed corresponding to the oil stain index is determined; the mathematical mapping relationship has the oil stain index as the independent variable and the fan compensation speed as the dependent variable, and the mathematical mapping relationship includes the range hood's preset speed and compensation coefficient; according to the fan compensation speed, the range hood's gear parameters for the next start-up are updated.

[0205] Based on the experimental relationship between oil pollution index and air volume compensation, the above mathematical mapping relationship can be constructed, which can be specifically expressed as:

[0206] ;

[0207] Where, N set The fan speed corresponding to the gear selected by the user. The compensation coefficient can be specifically calibrated through wind tunnel experiments, and its value range can be [0.1, 0.4]. Specifically, in this embodiment... It can take the value 0.35; N comp To compensate for the fan speed, it can be based on the calculated oil pollution index I. oil The corresponding fan compensation speed is used to set the fan speed parameters for the next startup. When the oil contamination index I... oil When the value is greater than 0.25, air volume compensation is activated, and the corresponding fan compensation speed is determined through the above mathematical mapping relationship.

[0208] Therefore, this embodiment can also automatically increase the fan speed according to the value of the oil stain index to compensate for the decrease in aerodynamic performance caused by oil stains. It solves the problem in related technologies that fixed-cycle maintenance cannot adapt to different usage intensities. It achieves the dual effects of accurate oil stain prediction and cleaning maintenance of the range hood, as well as adaptive compensation of performance, so that the cleaning cycle is extended by more than 30% while still having good oil fume suction power.

[0209] Figure 4 This is a schematic diagram of a signal flow in this embodiment, such as... Figure 4As shown, the camera and supplementary light form a sensing module that sends the acquired RGB images to the main controller. The main controller controls the supplementary light to provide supplementary lighting for the camera's image acquisition. During the control and processing phase, the main controller also performs image preprocessing and grayscale processing based on the received RGB images to obtain a target grayscale image. Then, it extracts oil stain feature regions from the target grayscale image and calculates the oil stain coverage rate. In addition, it extracts the grayscale distribution change index and multi-scale texture roughness from the target grayscale image. Then, during the decision and execution phase, the main controller calculates a comprehensive oil stain index based on the oil stain coverage rate, grayscale distribution change index, and multi-scale texture roughness, and maps the oil stain index to an oil stain level to achieve graded judgment. When the oil stain level is higher than L2, a graded cleaning reminder is sent to the mobile terminal and interactive display screen based on the user reminder unit. When the oil stain index is greater than or equal to 0.25, the fan speed compensation is activated, and a control command is sent to the fan drive module to control the fan speed level.

[0210] Figure 5 This is a flowchart of a multi-scale texture roughness extraction method according to this embodiment, such as... Figure 5 As shown, the multi-scale texture roughness extraction includes the following steps:

[0211] Step S501: Input the preprocessed target grayscale image;

[0212] Step S502: Perform level quantization on the grayscale of the target grayscale image;

[0213] Step S503: Based on the gray level quantization in step S502, calculate the gray co-occurrence matrix at each scale and angle.

[0214] Step S504: Normalize the gray-level co-occurrence matrix into a probability matrix;

[0215] Step S505: Calculate the roughness in each direction at a certain scale based on the probability matrix;

[0216] Step S506: Average the roughness in each direction at a certain scale to obtain the average roughness at that scale.

[0217] Step S507: Weighted fusion of the average roughness at each scale to obtain a comprehensive multi-scale texture roughness.

[0218] Figure 6 These are flowcharts of oil stain detection methods for range hoods in some embodiments, such as... Figure 6 As shown, the oil contamination status detection method may include the following steps:

[0219] Step S601: After a preset time delay following the shutdown of the range hood, start the oil stain status detection; for example, the preset time period can be 30 seconds.

[0220] Step S602: Obtain the original RGB image of the inside of the range hood;

[0221] Step S603: Preprocess and convert the RGB image to grayscale to obtain the target grayscale image; Steps S604 to S607 are executed respectively.

[0222] Step S604: Extract the characteristic areas of the oil stains;

[0223] Step S605: Calculate the oil pollution coverage rate based on the oil pollution characteristic areas; proceed to step S608.

[0224] Step S606: Calculate the grayscale distribution change index; proceed to step S608.

[0225] Step S607: Calculate the multi-scale texture roughness; proceed to step S608.

[0226] Step S608: Calculate the comprehensive oil pollution index;

[0227] Step S609: Determine whether the oil pollution level corresponding to the oil pollution index is greater than or equal to level 2; if yes, proceed to step S610; otherwise, proceed to step S612; level 2 here can correspond to level L2 mentioned above.

[0228] Step S610: Send a tiered cleaning reminder;

[0229] Step S611: Calculate airflow compensation and perform airflow compensation at different speeds before cleaning is complete; specifically, refer to the fan speed compensation in the above embodiment. Execute step S613;

[0230] Step S612: Record the data for this step;

[0231] Step S613: Update the next fan start-up gear parameters; end the process.

[0232] This embodiment also provides an oil stain status detection device for a range hood. This device is used to implement the above embodiments and preferred embodiments, and will not be repeated as already described. The terms "module," "unit," "subunit," etc., used below refer to combinations of software and / or hardware that achieve a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0233] Figure 7 This is a structural block diagram of the oil stain status detection device 70 of the range hood in this embodiment, as shown below. Figure 7As shown, the oil stain state detection device 70 of the range hood includes: an acquisition module 71, a feature extraction module 72, a weight determination module 73, and a weighted fusion module 74; wherein:

[0234] The acquisition module 71 is used to acquire a target grayscale image of the inside of the fan housing of the range hood when the range hood is turned off; the feature extraction module 72 is used to extract the oil stain coverage rate and grayscale distribution change index from the target grayscale image; and, according to the set scale and direction, extract the multi-scale texture roughness of the target grayscale image based on the grayscale co-occurrence matrix; the scale and direction are set based on the internal oil stain adhesion characteristics of the range hood; the weight determination module 73 is used to determine the first fusion weight corresponding to the oil stain coverage rate, the second fusion weight corresponding to the grayscale distribution change index, and the third fusion weight corresponding to the texture roughness based on the oil stain sample data covering different materials and / or structures, combined with the regression optimization algorithm; the weighted fusion module 74 is used to perform weighted fusion of the oil stain coverage rate, grayscale distribution change index, and texture roughness according to the first fusion weight, the second fusion weight, and the third fusion weight to obtain the oil stain index of the range hood.

[0235] It should be noted that the above modules can be functional modules or program modules, and can be implemented through software or hardware. For modules implemented through hardware, the above modules can reside in the same processor; or the above modules can be located in different processors in any combination.

[0236] This embodiment also provides a smart range hood, including a range hood body and an oil stain status detection device 70 for the range hood provided in the above embodiment. The smart range hood of this embodiment can accurately detect the oil stain status and provide cleaning reminders adapted to different usage habits and intensities, improving the timeliness of cleaning reminders, thereby enhancing the intelligence level of the range hood and improving the user experience.

[0237] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementations, and will not be repeated in this embodiment.

[0238] It should be understood that the specific embodiments described herein are merely illustrative of the application and not intended to limit it. All other embodiments derived by those skilled in the art based on the embodiments provided in this application without inventive effort are within the scope of protection of this application.

[0239] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0240] Obviously, the accompanying drawings are merely some examples or embodiments of this application. Those skilled in the art can apply this application to other similar situations based on these drawings without any creative effort. Furthermore, it is understood that although the work done in this development process may be complex and lengthy, for those skilled in the art, certain design, manufacturing, or production modifications made based on the technical content disclosed in this application are merely conventional technical means and should not be considered as insufficient disclosure of this application.

[0241] The term "embodiment" in this application refers to a specific feature, structure, or characteristic described in connection with an embodiment that may be included in at least one embodiment of this application. The appearance of this phrase in various places in the specification does not necessarily imply the same embodiment, nor does it imply that it is mutually exclusive with or independent of other embodiments. It will be clearly or implicitly understood by those skilled in the art that the embodiments described in this application may be combined with other embodiments without conflict.

[0242] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of patent protection. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the appended claims.

Claims

1. A method for detecting the oil stain status of a range hood, characterized in that, include: With the range hood turned off, acquire a target grayscale image of the inside of the fan housing of the range hood; The oil stain coverage rate and gray-level distribution change index are extracted from the target gray-level image; and, based on the set scale and direction, the multi-scale texture roughness of the target gray-level image is extracted based on the gray-level co-occurrence matrix. The dimensions and orientation are set based on the internal oil stain adhesion characteristics of the range hood; Based on oil stain sample data covering different materials and / or structures, and combined with regression optimization algorithms, the first fusion weight corresponding to the oil stain coverage rate, the second fusion weight corresponding to the gray scale distribution change index, and the third fusion weight corresponding to the texture roughness are determined. The oil stain coverage, the grayscale distribution change index, and the texture roughness are weighted and fused according to the first fusion weight, the second fusion weight, and the third fusion weight to obtain the oil stain index of the range hood.

2. The method for detecting the oil stain status of a range hood according to claim 1, characterized in that, Extracting the oil stain coverage rate and grayscale distribution change index from the target grayscale image includes: Based on the baseline clean state grayscale image, the oil stain feature region of the target grayscale image is determined, and the oil stain coverage rate is extracted based on the oil stain feature region; Calculate the first gray-level histogram of the baseline clean state gray-level image and the second gray-level histogram of the target gray-level image respectively. Based on the first gray-level histogram and the second gray-level histogram, determine the gray-level distribution change index of the target gray-level image.

3. The method for detecting the oil stain status of a range hood according to claim 2, characterized in that, Based on a baseline clean state grayscale image, the oil stain feature regions of the target grayscale image are determined, and the oil stain coverage is extracted based on the oil stain feature regions, including: Calculate the pixel color difference image between the target grayscale image and the baseline clean state grayscale image; Based on the inter-class variance, an adaptive segmentation threshold for the pixel color difference image is determined; Based on the adaptive segmentation threshold, the oil stain feature region is segmented from the pixel color difference image; The oil stain coverage rate is determined by the ratio between the number of pixels in the oil stain feature area and the total number of pixels in the pixel color difference image.

4. The method for detecting the oil stain status of a range hood according to claim 2, characterized in that, Based on the first gray-level histogram and the second gray-level histogram, the gray-level distribution change index of the target gray-level image is determined, including: Calculate the chi-square distance between the first gray-level histogram and the second gray-level histogram, and the rate of change of the mean gray level between the first gray-level histogram and the second gray-level histogram; The gray-scale distribution change index is obtained by weighting and fusing the chi-square distance and the gray-scale mean change rate.

5. The method for detecting the oil stain status of a range hood according to claim 1, characterized in that, Based on the set scale and direction, the multi-scale texture roughness of the target grayscale image is extracted using the gray-level co-occurrence matrix, including: Based on the physical size characteristics of oil stains on the impeller surface of range hoods collected in history, microscale, mesoscale, and macroscale are set for gray-level co-occurrence matrix calculation; based on the distribution characteristics of oil stains inside range hoods collected in history, multiple directions are set for gray-level co-occurrence matrix calculation. For the set microscale, mesoscale, and macroscale, calculate the corresponding gray-level co-occurrence matrix under each set direction; All gray-level co-occurrence matrices are fused to obtain multi-scale texture roughness.

6. The method for detecting the oil stain status of a range hood according to claim 5, characterized in that, By fusing all gray-level co-occurrence matrices, multi-scale texture roughness is obtained, including: For each of the set scales, the average roughness of the set multiple directions is calculated based on the probability matrix of the gray-level co-occurrence matrix corresponding to each direction. The average roughness of each scale is weighted and fused to obtain multi-scale texture roughness.

7. The method for detecting the oil stain status of a range hood according to claim 1, characterized in that, Based on oil stain sample data covering different materials and / or structures, and combined with a regression optimization algorithm, the first fusion weight corresponding to the oil stain coverage rate, the second fusion weight corresponding to the grayscale distribution change index, and the third fusion weight corresponding to the texture roughness are determined, including: Based on historically collected grayscale images of internal samples of range hoods with different materials and / or structures, combined with labeled oil stain level tags, oil stain sample data is constructed; the sample grayscale images include several sets of images, each set of images being grayscale images of the same range hood sample under different oil stain levels, and the materials and / or structures of the range hood samples in different sets of images are different. Based on the oil pollution sample data and the ridge regression optimization algorithm, the first fusion weight, the second fusion weight, and the third fusion weight are optimized to obtain the optimized first fusion weight, the second fusion weight, and the third fusion weight.

8. The method for detecting the oil stain status of a range hood according to claim 1, characterized in that, After obtaining the oil pollution index, the method further includes: Based on the calibrated mathematical mapping relationship between the oil stain index and air volume compensation, the fan compensation speed corresponding to the oil stain index is determined; the mathematical mapping relationship uses the oil stain index as the independent variable and the fan compensation speed as the dependent variable, and the mathematical mapping relationship includes the range hood preset speed and compensation coefficient. The range hood's speed setting for the next startup is updated based on the fan's compensated rotation speed.

9. A device for detecting the oil stain status of a range hood, characterized in that, include: The module comprises an acquisition module, a feature extraction module, a weight determination module, and a weighted fusion module; among which: The acquisition module is used to acquire a target grayscale image inside the fan housing of the range hood when the range hood is turned off. The feature extraction module is used to extract the oil stain coverage rate and gray-level distribution change index from the target gray-level image; and, based on the gray-level co-occurrence matrix, extract the multi-scale texture roughness of the target gray-level image according to the set scale and direction; the scale and direction are set based on the internal oil stain adhesion characteristics of the range hood. The weight determination module is used to determine, based on oil stain sample data covering different materials and / or structures, and in conjunction with a regression optimization algorithm, the first fusion weight corresponding to the oil stain coverage rate, the second fusion weight corresponding to the grayscale distribution change index, and the third fusion weight corresponding to the texture roughness. The weighted fusion module is used to perform weighted fusion of the oil stain coverage, the gray scale distribution change index and the texture roughness according to the first fusion weight, the second fusion weight and the third fusion weight to obtain the oil stain index of the range hood.

10. A smart range hood, characterized in that, It includes the range hood body and the oil stain status detection device for the range hood as described in claim 9.

Images