Fragrance release device and method based on three-dimensional machine vision and sound perception

Abstract

The invention relates to the technical field of fragrance controlled release, and provides a fragrance release device and method based on three-dimensional machine vision and sound perception.The fragrance release device comprises a vision sensor, a sound sensor, an atomization head, an atmosphere lamp and an analysis control module; the visual sensor is used for acquiring an environment dominant hue and object information of an environment where a user is located currently, the sound sensor is used for acquiring voice information of the environment where the user is located currently, the atomizing head is used for releasing fragrance, and the atmosphere lamp is used for forming lamplight for adjusting the atmosphere; and the analysis control module is used for analyzing at least one piece of environment information of the environment dominant hue, the object information and the voice information of the current environment of the user to obtain a control instruction, and controlling the atomization head and / or the atmosphere lamp to adjust at least one of the fragrance type, the fragrance concentration and / or the atmosphere lamp color and the atmosphere lamp brightness according to the control instruction.

Description

Technical Field

[0001] The present invention relates to the technical field of fragrance controlled release, in particular to a fragrance release device and method based on three-dimensional machine vision and sound perception. Background Art

[0002] Existing traditional fragrance management systems usually rely on timing or manual control to release fragrance. They are unable to make intelligent adjustments based on the user's real-time needs and lack the ability to identify the user's status, resulting in a fragrance experience that is not personalized and flexible enough. Although sensor-based environmental monitoring systems can obtain environmental data through sensors such as temperature and humidity, most of them cannot fully analyze the user's behavior and emotions, and lack a real-time feedback mechanism. They cannot adjust the type and intensity of fragrance in time according to changes in the user's status. Therefore, existing devices cannot provide refined personalized services. Summary of the invention

[0003] In view of the above-mentioned defects, the purpose of the present invention is to propose a fragrance release device and method based on three-dimensional machine vision and sound perception, aiming to adjust the release of fragrance based on the ambient color tone and the user's voice command requirements.

[0004] To achieve this object, the present invention adopts the following technical solutions:

[0005] A fragrance release device based on three-dimensional machine vision and sound perception, including a visual sensor, a sound sensor, an atomizer head, an atmosphere light, and an analysis and control module;

[0006] The visual sensor is used to obtain the main color tone and object information of the environment in which the user is currently located, the sound sensor is used to obtain the voice information of the environment in which the user is currently located, the atomizer head is used to release fragrance, and the atmosphere light is used to form a light to adjust the atmosphere;

[0007] The analysis and control module is used to analyze at least one of the environmental information including the main color tone of the environment, object information and voice information of the user's current environment to obtain control instructions, and control the atomizer head and / or the atmosphere light to adjust at least one of the fragrance type, fragrance concentration and / or atmosphere light color, and atmosphere light brightness according to the control instructions.

[0008] Preferably, the visual sensor acquires the main color tone of the environment in which the user is currently located, including:

[0009] The visual sensor acquires a color image of the environment where the user is located, and converts the color image of the environment where the user is located from an RGB color space to an HSV color space;

[0010] A color histogram is obtained by calculating the frequency distribution of each color in the color image, and pixels in each color quantization interval of the color histogram are divided into aggregated pixels and non-aggregated pixels;

[0011] The color correlation between pixels is analyzed to obtain a color correlation map, and the main color tone of the environment is detected by combining the color histogram and the color correlation map.

[0012] Preferably, the object information includes user gesture;

[0013] The visual sensor acquires the user posture of the user's current environment including:

[0014] Obtain 3D point cloud data of the user's current environment and use the human posture estimation algorithm to detect the key points of the human body in the 3D point cloud;

[0015] Constructing a human skeleton model according to the key points, connecting each key point to form a joint;

[0016] Analyze the user's posture type based on the angles and relative positions between each joint, and classify the user's posture;

[0017] At least one of the preset fragrance type, fragrance concentration, ambient light color and ambient light brightness is matched according to the classification result.

[0018] Preferably, the object information includes the number of users;

[0019] The visual sensor obtains the number of users in the current environment of the user including:

[0020] Obtain 3D point cloud data and 2D images of the user's current environment;

[0021] Perform image segmentation on the 2D image, separate the user from the background in the 2D image, and identify and extract the outline of each user based on the depth information;

[0022] Extract the features of each user and calculate the number of users in the environment based on the profile and features of each user;

[0023] At least one of the preset fragrance type, fragrance concentration, ambient light color and ambient light brightness is matched according to the number of users.

[0024] Preferably, the voice information includes user voice instructions;

[0025] The sound sensor obtains the voice command of the user's current environment including:

[0026] Extract features from the user's voice commands and set a number of corresponding thresholds for each feature;

[0027] Comparing each feature with a corresponding threshold to identify silent parts and non-silent parts in the user's voice command;

[0028] After smoothing the user's voice command, the silent part is removed to obtain several valid voice segments.

[0029] Furthermore, the sound sensor further includes:

[0030] Use a speech recognition engine to convert valid speech segments into text information;

[0031] identifying the first fragrance keyword in the text information according to the keyword extraction algorithm;

[0032] The text information is classified according to the preset classification algorithm to obtain several intent classification labels;

[0033] A second fragrance keyword in the text information is identified based on the first keyword, and a command key sentence is obtained according to the intention classification label and the second fragrance keyword to form a user voice command.

[0034] Furthermore, if there are voice commands from multiple users, the voice commands of each user are obtained respectively, and the command key sentences corresponding to each user are obtained to form the voice commands of each user. After a preset length of time, the fragrance type, fragrance concentration, ambient light color and ambient light brightness are adjusted based on the voice command of one of the users.

[0035] Preferably, the voice information also includes environmental voice.

[0036] A fragrance release method based on a three-dimensional visual sensor is applied to a fragrance release device including an atmosphere lamp and an atomizer head, wherein the atomizer head is used to release fragrance, and the atmosphere lamp is used to form a light for adjusting the atmosphere. The fragrance release method includes:

[0037] Get the main color tone and object information of the user's current environment;

[0038] Get the voice information of the user's current environment;

[0039] At least one of the main color tone of the environment, object information and voice information of the user's current environment is analyzed to obtain a control instruction, and at least one of the fragrance type, fragrance concentration, ambient light color and ambient light brightness is adjusted according to the control instruction.

[0040] One of the above technical solutions has the following advantages or beneficial effects:

[0041] The present invention utilizes visual sensors to identify the main color of the user's environment in real time, and adjusts the fragrance type and the color of the atmosphere light according to the emotional characteristics of the main color. It can adapt to different environmental atmospheres and provide fragrance and lighting effects that meet the situation, greatly improving the user's immersive experience. It accurately identifies the user's effective voice commands through sound sensors to avoid interference from background noise, which helps to ensure the accuracy of voice recognition. By analyzing the commands and making secondary adjustments to the fragrance type and concentration, and the color and brightness of the atmosphere light, the user only needs to issue simple commands, such as "adjust the fragrance concentration to medium" or "dim the lights", to automatically adjust the corresponding device settings, thereby improving the convenience and intelligence of user interaction. Through comprehensive perception of the environment and voice control, the device settings are adjusted in a timely manner according to user feedback to enhance the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0043] Figure 1 is a schematic diagram of the structure of a fragrance release device based on three-dimensional machine vision and sound perception provided by an embodiment of the present invention;

[0044] Figure 2 is a flow chart of a fragrance release method based on three-dimensional machine vision and sound perception provided by an embodiment of the present invention;

[0045] Among them, there are an atomizer head 1, an atmosphere light 2, a visual sensor and sound sensor module 3, and an operation panel 4. DETAILED DESCRIPTION

[0046] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0047] In this application, the terms "comprises", "comprising" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprising a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element.

[0048] Existing traditional fragrance management systems usually rely on timing or manual control to release fragrance. They are unable to make intelligent adjustments based on the user's real-time needs and lack the ability to identify the user's status, resulting in a fragrance experience that is not personalized and flexible enough. Although sensor-based environmental monitoring systems can obtain environmental data through sensors such as temperature and humidity, most of them cannot fully analyze the user's behavior and emotions, and lack a real-time feedback mechanism. They cannot adjust the type and intensity of fragrance in time according to changes in the user's status. Therefore, existing devices cannot provide refined personalized services.

[0049] like Figure 1 As shown, a three-dimensional visual machine perception fragrance device includes a visual sensor, a sound sensor, an atomizer head 1, an atmosphere light 2 and an analysis control module;

[0050] The visual sensor and the sound sensor constitute the visual sensor and sound sensor module 3, wherein the visual sensor is used to obtain the main color tone and object information of the environment in which the user is currently located, the sound sensor is used to obtain the voice information of the environment in which the user is currently located, the atomizer head is used to release fragrance, and the atmosphere light is used to form a light to adjust the atmosphere;

[0051] The analysis and control module is used to analyze at least one of the main color tone of the environment, object information and voice information of the user's current environment to obtain a control instruction, and adjust at least one of the fragrance type, fragrance concentration, ambient light color and ambient light brightness according to the control instruction.

[0052] In an optional embodiment, a visual sensor (usually a camera or RGB sensor) is used to capture images and color information of the user's environment. Through image analysis technology, the device can identify the main color, objects (such as furniture, wall color, etc.) and even the layout of the room in the environment, which helps to judge the overall visual atmosphere of the space (such as warmth, calmness, activity, etc.). For example, if a warm red tone is detected, the device may infer that the environment needs a more relaxing and soothing fragrance. Sound sensors (such as microphones) are used to collect surrounding sounds, especially voice information. These sounds can be user conversations, sound effects of the surrounding environment, or changes in other sounds. The sound sensor inputs the captured audio data into the analysis and control module, and the device performs emotional analysis to identify the emotional characteristics of the voice (such as happiness, anxiety, sadness, etc.). For example, when negative emotions (such as low or hurried tone) are detected in the voice, it can automatically adjust the fragrance type (such as using lavender fragrance) and reduce the light brightness to create a more soothing environment. The atomizer head is the core component used to release the fragrance. It converts liquid fragrance into tiny fragrance particles through ultrasound or other atomization technology and evenly distributes them in the air. Through the connection with the analysis and control module, the atomizer head can automatically adjust the type and concentration of fragrance according to the command. For example, if the analysis and control module determines that the user is in a relaxed state based on the input of the visual and sound sensors, it will select a gentle floral or citrus fragrance and adjust the concentration to a low level to avoid over-stimulation. Figure 1The aroma release device shown in the figure has essential oil inside its atomizer head. The aroma release device is equipped with an ultrasonic vibrator, which uses the high-frequency oscillation of ultrasound to decompose water molecules and dissolved plant essential oils into nano-scale cold mist with a diameter of 0.1-5 microns, and then emits it into the surrounding air. The concentration is determined by changing the power of ultrasonic vibration. The ambient light provides visual atmosphere adjustment and can affect the emotional atmosphere of the space by changing the color and brightness. The adjustment of the ambient light is based on the output of the analysis and control module, and the color and brightness will automatically change according to the needs of the environment. For example, in a more depressing environment, the ambient light will automatically be adjusted to a warm tone (such as soft yellow or orange) and the brightness will be reduced to make the environment more warm and comfortable. In a fresh environment, cold-toned lights, such as blue or purple, will be selected to enhance the refreshing feeling of the space. Different lights can also be emitted at the same time. The analysis and control module is responsible for comprehensive processing of input data from visual sensors and sound sensors. It analyzes environmental information based on a preset algorithm and emotion recognition model to judge the user's emotional state, environmental atmosphere, and visual style. For example, based on the main color in the image and the emotional characteristics in the sound, it is determined whether the current environment is suitable for relaxation, activity or concentration. Based on the analysis results, the module issues a control instruction to adjust the fragrance type and concentration of the atomizer head, and the color and brightness of the atmosphere light. The user does not need to operate manually. The device proposed in the present invention can achieve dynamic adaptive adjustment according to environmental changes or emotional states to enhance the user's physical and mental experience. In addition, the fragrance release device proposed in the present invention is also provided with an operation panel 4. When the user is not satisfied with the adjustment of the fragrance release device, the user can manually adjust the release of the fragrance and the brightness and color of the atmosphere light.

[0053] Preferably, the visual sensor acquires the main color tone of the environment in which the user is currently located, including:

[0054] The visual sensor acquires a color image of the environment where the user is located, and converts the color image of the environment where the user is located from an RGB color space to an HSV color space;

[0055] A color histogram is obtained by calculating the frequency distribution of each color in the color image, and pixels in each color quantization interval of the color histogram are divided into aggregated pixels and non-aggregated pixels;

[0056] The color correlation between pixels is analyzed to obtain a color correlation map, and the main color tone of the environment is detected by combining the color histogram and the color correlation map.

[0057] RGB (Red, Green, and Blue) and HSV (Hue, Saturation, and Value) are two commonly used color representation methods. The RGB color space defines colors by the intensities of the three channels of red, green, and blue, while the HSV color space represents colors by hue, saturation, and value. The HSV model is more in line with human visual perception than RGB because it separates the hue, purity, and brightness of colors for intuitive understanding. By converting RGB to HSV, the influence of lighting can be eliminated, making the color representation more robust. Especially when identifying the environmental color tone, HSV can more accurately reflect the main color tone of the environment than RGB.

[0058] The color histogram is a statistical tool used to represent the distribution of different colors in an image. By calculating the frequency of each color in the image, a distribution diagram reflecting the overall hue and main color of the image can be obtained. The frequency of each color reflects the number of times the color appears in the image. The implementation method can be to count the color of each pixel to obtain the number of times or frequency of each color in the image. The X-axis of the histogram is the color range (usually divided by hue, saturation and brightness), and the Y-axis is the number of times the corresponding color appears or the number of pixels. The color space is then divided into multiple intervals (for example, divided into several segments by hue, each segment includes a color range), and then the frequency of the color in each interval is counted. Through the color histogram, the most important color distribution in the image can be obtained very intuitively, which is helpful for the subsequent identification of the main color.

[0059] The pixels in each color quantization interval in the image are divided into two categories: aggregated pixels and non-aggregated pixels. Aggregated pixels refer to pixels with higher color frequency and relatively concentrated in space; non-aggregated pixels refer to pixels with lower color frequency or sparse distribution. The distinction operation can be to set a color frequency threshold. Colors exceeding the threshold are considered aggregated colors, and colors below the threshold are considered non-aggregated colors. In this way, unimportant colors and noise can be filtered out. Clustering algorithms (such as K-means) are used to cluster colors, and pixels of similar colors are clustered together to form color aggregation areas. Then, according to the color frequency of each area, they are divided into aggregated and non-aggregated categories. By distinguishing aggregated pixels and non-aggregated pixels, background noise, redundant colors or irrelevant color information can be effectively eliminated. By analyzing the spatial distribution of high-frequency color areas in the image, the main color features are extracted and the main color area is identified. In an image, adjacent or spatially close pixels usually have similar colors. Color correlation refers to the similarity between the colors of these adjacent pixels, which can be evaluated by calculating the distance between color values ​​(such as Euclidean distance). The color correlation diagram describes the distribution of colors in space by analyzing the color similarity between pixels in the image. It can reveal which colors are interrelated in the image and which colors are continuous in space. The color correlation diagram not only shows the similarity between colors, but also reveals the spatial distribution pattern of colors in the image, which helps to understand the overall structure of the image. By analyzing the spatial relationship between colors, it can avoid misidentification that relies solely on frequency distribution, especially when there are multiple colors or complex color distribution in the image.

[0060] The color histogram reflects the frequency distribution of each color in the image, while the color correlation diagram reflects the spatial distribution and correlation of colors in the image. Combining the two, the main color tone of the image can be determined from both the frequency and spatial structure of the color.

[0061] Preferably, the object information includes user gesture;

[0062] The visual sensor acquires the user posture of the user's current environment including:

[0063] Obtain 3D point cloud data of the user's current environment and use the human posture estimation algorithm to detect the key points of the human body in the 3D point cloud;

[0064] Constructing a human skeleton model according to the key points, connecting each key point to form a joint;

[0065] The user posture type is analyzed based on the angles and relative positions between each joint, and the user posture is classified.

[0066] In this embodiment, the 3D point cloud data in the user's environment is first obtained through a visual sensor. These data contain the three-dimensional coordinate information of the user in space. The key points of the human body are detected from the 3D point cloud using a human posture estimation algorithm, which may include the positions of important parts such as the head, shoulders, elbows, wrists, knees, ankles, etc. According to the detected key points, a human skeleton model is formed by connecting these key points. Each key point represents a joint of the human body, and these joints are connected to each other through lines to form a dynamic skeleton structure. According to the angles and relative positions between the joints in the skeleton model, the user's posture type is analyzed. For example, it can be determined whether the user is standing, sitting, walking, bending over or other specific postures. Based on these analyses, the device can classify the user's posture and identify its behavior state or action type. After recognizing the user's current posture, the preset fragrance type, fragrance concentration, ambient light color and ambient light brightness are automatically matched, and then the atomizer head or fragrance light is controlled to perform the matched preset operation. For example, the visual sensor recognizes that the user is lying flat, and the preset operation matched to the lying posture is to start the lavender fragrance and adjust the fragrance concentration to a medium level, while adjusting the ambient light to a soft yellow and reducing the brightness to 40%. At this point, the lighting and fragrance concentration in the room are perfectly matched to the user's needs, creating an ideal resting environment. When the visual sensor recognizes that the user is standing, the preset operation matched to the standing posture is that the ambient light will automatically switch from the low-brightness warm light of the night mode to the bright cold light of the day mode. For example, after the user gets up in the morning, citrus fragrance will be automatically sprayed to create a refreshing atmosphere, and at night it will automatically switch to lavender fragrance to help the user relax and fall asleep. This intelligent automatic adjustment mechanism does not require manual intervention by the user, ensuring that each environmental setting can automatically adapt according to time, air quality and user behavior, providing a comfortable living space at all times.

[0067] Preferably, the object information includes the number of users;

[0068] The visual sensor obtains the number of users in the current environment of the user including:

[0069] Obtain 3D point cloud data and 2D images of the user's current environment;

[0070] Perform image segmentation on the 2D image, separate the user from the background in the 2D image, and identify and extract the outline of each user based on the depth information;

[0071] Extract the features of each user and calculate the number of users in the environment based on the profile and features of each user.

[0072] The visual sensor first captures the spatial information of the current environment through 3D point cloud data and 2D images. The 3D point cloud data provides the depth information of the scene, while the 2D image helps capture the visual details in the image. The combination of the two can fully understand the location and characteristics of objects in the environment. When processing 2D images, image segmentation is first performed to separate the user from the background in the image. This can be achieved through deep learning models (such as convolutional neural networks). The model can identify and separate the difference between users and other objects (such as furniture, walls, etc.) by analyzing features such as color, shape and texture of the image. Once the user in the image is separated, the user area in the 2D image can be further processed in combination with the depth information to accurately identify and extract the outline of each user. The depth information usually comes from 3D point cloud data. Through the analysis of the point cloud, the position and outline of the user can be accurately located. For each extracted user outline, the visual sensor will further extract the user's features, such as body shape, facial features, posture, etc., which not only helps to further verify the user's identity, but also can be used to distinguish different users, especially in multi-person environments, to ensure that the system can accurately identify each user. Based on the outline and features of each user, the visual sensor will count and calculate the number of users in the current environment. It combines the results of outline recognition and feature extraction, and uses an algorithm to judge each independent user object to derive the final number of users.

[0073] After obtaining the number of users in the current environment of the device, the corresponding fragrance type, fragrance concentration, atmosphere light color and atmosphere light brightness are automatically matched and adjusted according to the number of users, and then the atomizer head or fragrance light is controlled to perform the corresponding operation. In one embodiment, for example, if the number of users matched is multiple people, which belongs to a group state, then the operation performed by the atomizer head and the fragrance light can be to increase the concentration of the released fragrance and to increase the brightness of the atmosphere light. If the number of users is only one, then the concentration of the released fragrance can be lowered, etc.; in another embodiment, the number of users can also be combined with other environmental information such as user posture, the main color of the environment, and voice information to more accurately control and adjust the atomizer head and fragrance light; for example, if it is detected that the current number of users is one, and the main color of the environment is blue, then the matched preset operation is that the atomizer head will release a relaxing aroma.

[0074] Preferably, the voice information includes a user voice command. It should be noted that, in this embodiment, the voice command refers to a command issued by the user to explicitly control the atomizer head or the fragrance lamp of the device, such as "turn on the atomizer head", "release lavender fragrance", "turn on the fragrance lamp", "adjust the brightness to level 1", and other explicit commands;

[0075] The sound sensor obtains the voice command of the user's current environment including:

[0076] Extract features from the user's voice commands and set a number of corresponding thresholds for each feature;

[0077] Comparing each feature with a corresponding threshold to identify silent parts and non-silent parts in the user's voice command;

[0078] After smoothing the user's voice command, the silent part is removed to obtain several valid voice segments.

[0079] Among them, feature extraction is used to describe different aspects of speech signals, so as to more effectively distinguish between silence and non-silence. The extracted features include short-time energy (STE): a measure of signal strength, which can effectively distinguish between speech and silence; zero-crossing rate (ZCR): describing the frequency of change of signal waveform, which can reflect the characteristics of speech signals; Mel frequency cepstral coefficients (MFCC): used for speech recognition, which can capture the spectral characteristics of speech. For each feature (such as short-time energy, zero-crossing rate, etc.), it is necessary to set a suitable threshold. The selection of the threshold can be based on experience or through training data tuning to ensure that silence and non-silence can be effectively distinguished. For example, for the short-time energy threshold, a threshold can be set according to the strength of the signal. When the energy value of the signal is lower than the threshold, it is considered to be silence or noise. By comparing each feature with the corresponding threshold, it is possible to identify which parts are silence and which parts are valid parts with speech. In addition, it can be processed by sliding a certain time window to capture the changing audio signal. In actual speech signals, the division between silence and non-silence may be affected by noise or other factors, resulting in an unclear or inaccurate boundary. Therefore, smoothing the signal helps to remove errors caused by noise or other factors, making the division of silence segments and valid speech segments more accurate and coherent. Through smoothing processing (such as sliding average), the start and end positions of the speech segments can be more accurately identified. After smoothing, all parts identified as silence can be removed and only valid speech segments are retained.

[0080] Preferably, the sound sensor further comprises:

[0081] Use a speech recognition engine to convert valid speech segments into text information;

[0082] identifying the first fragrance keyword in the text information according to the keyword extraction algorithm;

[0083] The text information is classified according to the preset classification algorithm to obtain several intent classification labels;

[0084] A second fragrance keyword in the text information is identified based on the first keyword, and a command key sentence is obtained according to the intention classification label and the second fragrance keyword to form a user voice command.

[0085] A speech recognition engine (Microsoft Azure Speech) can be used to convert the user's voice instructions into text, where the keyword extraction algorithm can use text processing techniques (such as TF-IDF, TextRank, word embedding models, etc.) to identify keywords from the converted text. In this scenario, specific fragrance-related words (such as "fragrance", "concentration", "lighting", etc.) are identified as important keywords. Then, a trained classification algorithm (such as support vector machine, deep learning model, etc.) is applied to classify the text intent to determine the user's needs. For example, the intent of "releasing the smell of the ocean" is identified as "fragrance type setting". If a command contains multiple intents (such as adjusting the fragrance and lighting at the same time), a multi-label classification method can be used for processing. Based on the first fragrance keyword (such as "fragrance", "concentration", "lighting"), the text is further analyzed to find the second fragrance keyword, such as fragrance type ("ocean", "floral"), concentration ("increase", "decrease"), and lighting settings ("brighten", "dim"), and the second fragrance keyword can be extracted and classified using the NER model. Then, based on the extracted keywords and intentions, the natural language instructions input by the user are converted into a structured data format to obtain the key sentences of the instructions, such as converting "make the fragrance stronger" into (fragrance / concentration / increase").

[0086] Furthermore, if there are voice commands from multiple users, the voice commands of each user are obtained respectively, and the command key sentences corresponding to each user are obtained to form the voice commands of each user. After a preset length of time, the fragrance type, fragrance concentration, ambient light color and ambient light brightness are adjusted based on the voice command of one of the users.

[0087] When receiving voice commands from multiple users, step S2 will be executed for each user's voice command respectively to extract each user's command key sentence, such as fragrance type, fragrance concentration, ambient light color and brightness, etc. Then, after a preset time interval (such as every 5 seconds or 10 seconds), based on the command of one of the users, the device settings will be adjusted in turn, such as adjusting the type and concentration of the fragrance, the color and brightness of the ambient light. The commands of multiple users can be processed in turn or selected according to priority to ensure that each user's needs can be responded to, and the needs of multiple users can be efficiently managed, and precise control of the device can be achieved.

[0088] Preferably, the voice information also includes environmental voice. It should be noted that, in this embodiment, the environmental voice refers to other environmental sounds other than the non-explicit user command sound that can be obtained within a certain range where the device is currently located.

[0089] For example, the sound sensor acquires ambient voice, such as sad music played on TV. By performing emotional analysis on the sound of the TV, the fragrance type, concentration, brightness and color of the ambient light can be automatically adjusted. For example, when the system recognizes sad emotions, it will select a soothing fragrance (such as lavender) and set its concentration to medium to help ease the mood. At the same time, the ambient light will be adjusted to a soft blue or purple, and the brightness will be reduced to 30%-50%, creating a quiet and comfortable environment. This is done automatically by the device, and the user does not need to operate manually, providing the user with a comfortable environment.

[0090] A fragrance release method based on a three-dimensional visual sensor is applied to a fragrance release device including an atmosphere lamp and an atomizer head, wherein the atomizer head is used to release fragrance, and the atmosphere lamp is used to form a light for adjusting the atmosphere. The fragrance release method includes:

[0091] Get the main color tone and object information of the user's current environment;

[0092] Get the voice information of the user's current environment;

[0093] At least one of the main color tone of the environment, object information and voice information of the user's current environment is analyzed to obtain a control instruction, and at least one of the fragrance type, fragrance concentration, ambient light color and ambient light brightness is adjusted according to the control instruction.

[0094] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0095] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A fragrance release device based on three-dimensional machine vision and sound perception, characterized in that: Includes visual sensors, sound sensors, atomizer heads, ambient lights and analysis control modules; The visual sensor is used to obtain the main color tone and object information of the environment in which the user is currently located, the sound sensor is used to obtain the voice information of the environment in which the user is currently located, the atomizer head is used to release fragrance, and the atmosphere light is used to form a light to adjust the atmosphere; The analysis and control module is used to analyze at least one of the environmental information including the main color tone of the environment, object information and voice information of the user's current environment to obtain control instructions, and control the atomizer head and / or the atmosphere light to adjust at least one of the fragrance type, fragrance concentration and / or atmosphere light color, and atmosphere light brightness according to the control instructions.

2. The fragrance release device according to claim 1, characterized in that: The visual sensor obtains the main color tone of the environment where the user is currently located, including: The visual sensor acquires a color image of the environment where the user is located, and converts the color image of the environment where the user is located from an RGB color space to an HSV color space; A color histogram is obtained by calculating the frequency distribution of each color in the color image, and pixels in each color quantization interval of the color histogram are divided into aggregated pixels and non-aggregated pixels; The color correlation between pixels is analyzed to obtain a color correlation map, and the main color tone of the environment is detected by combining the color histogram and the color correlation map.

3. The fragrance release device according to claim 1, characterized in that: Object information includes user posture; The visual sensor acquires the user posture of the user's current environment including: Obtain 3D point cloud data of the user's current environment and use the human posture estimation algorithm to detect the key points of the human body in the 3D point cloud; Constructing a human skeleton model according to the key points, connecting each key point to form a joint; Analyze the user's posture type based on the angles and relative positions between each joint, and classify the user's posture; At least one of the preset fragrance type, fragrance concentration, ambient light color and ambient light brightness is matched according to the classification result.

4. The fragrance release device according to claim 1, characterized in that: The object information includes the number of users; The visual sensor obtains the number of users in the current environment of the user including: Obtain 3D point cloud data and 2D images of the user's current environment; Perform image segmentation on the 2D image, separate the user from the background in the 2D image, and identify and extract the outline of each user based on the depth information; Extract the features of each user and calculate the number of users in the environment based on the profile and features of each user; At least one of the preset fragrance type, fragrance concentration, ambient light color and ambient light brightness is matched according to the number of users.

5. The fragrance release device according to claim 1, characterized in that: The voice information includes a user voice command; The sound sensor obtains the voice command of the user's current environment including: Extract features from the user's voice commands and set a number of corresponding thresholds for each feature; Comparing each feature with a corresponding threshold to identify silent parts and non-silent parts in the user's voice command; After smoothing the user's voice command, the silent part is removed to obtain several valid voice segments.

6. The fragrance release device according to claim 5, characterized in that: The sound sensor further includes: Use a speech recognition engine to convert valid speech segments into text information; identifying the first fragrance keyword in the text information according to the keyword extraction algorithm; The text information is classified according to the preset classification algorithm to obtain several intent classification labels; A second fragrance keyword in the text information is identified based on the first keyword, and a command key sentence is obtained according to the intention classification label and the second fragrance keyword to form a user voice command.

7. The fragrance release device according to claim 6, characterized in that: If there are voice commands of multiple users, the voice commands of each user are obtained respectively, and the key command sentences corresponding to each user are obtained to form the voice commands of each user. After a preset length of time, the fragrance type, fragrance concentration, ambient light color and ambient light brightness are adjusted based on the voice command of one of the users.

8. The fragrance release device according to claim 1, characterized in that: The voice information also includes environmental voice.

9. A fragrance release method based on a three-dimensional visual sensor, applied to a fragrance release device including an atmosphere lamp and an atomizer head, wherein the atomizer head is used to release fragrance, and the atmosphere lamp is used to form a light for adjusting the atmosphere, characterized in that: The fragrance releasing method comprises: Get the main color tone and object information of the user's current environment; Get the voice information of the user's current environment; At least one of the main color tone of the environment, object information and voice information of the user's current environment is analyzed to obtain a control instruction, and at least one of the fragrance type, fragrance concentration, ambient light color and ambient light brightness is adjusted according to the control instruction.

Images