Intelligent garbage detection method based on light source visual lens recognition

Abstract

The invention relates to the technical field of garbage classification treatment, and discloses an intelligent garbage detection method based on light source visual lens recognition, and the method comprises the following steps: S1, capturing garbage and environment image data through a visual sensor; s2, acquiring an image by using a quantum optical enhanced visual lens system; s3, a CANDECOMPP / PARAFAC method is adopted to convert the image data into a low-dimensional matrix; s4, inputting the low-dimensional matrix into a deep neural network for garbage recognition and classification; and S5, a robot path is calculated through a path optimization algorithm, and a garbage grabbing task is completed. Through cooperation of the quantum superposition state light source and the quantum entanglement state light source, enhancement processing of garbage image data is realized, so that definition and detail resolution of the image are improved, interference of environmental noise is effectively reduced, details and edges of garbage objects are more obvious, and improvement of garbage classification precision is promoted.

Description

Technical Field

[0001] The present invention relates to the technical field of garbage classification and treatment, and specifically to an intelligent garbage detection method based on light source vision lens recognition. Background Art

[0002] Existing garbage classification usually relies on computer vision and traditional machine learning algorithms to identify and classify garbage through image processing technology. It generally uses ordinary vision sensors or cameras to obtain environmental images and relies on methods such as image enhancement and feature extraction for classification. Traditional methods rely on manual selection of features or use classical image processing techniques such as edge detection and color extraction. These methods have improved the accuracy of garbage classification to a certain extent. At the same time, some more advanced garbage classification systems have also begun to use deep learning, especially convolutional neural networks, to automatically learn the features of garbage objects, thereby improving classification accuracy. However, despite the continuous progress of these technologies, the existing technologies still face some limitations, which affect the overall performance of the system.

[0003] First of all, the image quality of ordinary vision sensors in complex environments cannot meet the requirements of accurate classification. Especially in low-light environments and high-noise scenes, the image details are often blurred, affecting the recognition accuracy of garbage objects. Secondly, although the existing machine learning and deep learning methods have certain improvements in classification accuracy, they still face the problems of low computational efficiency and long training time. Especially when dealing with large-scale garbage images, the training process is very slow and difficult to meet the requirements of real-time classification. In addition, the application of existing path planning algorithms in garbage collection robots mainly relies on static maps and lacks the ability to adapt to dynamic environments in real time, resulting in the robot being unable to effectively respond to obstacles and environmental changes. These problems limit the further development of garbage classification systems in terms of automation and intelligence. Summary of the Invention

[0004] Aiming at the deficiencies of the existing technology, the present invention provides an intelligent garbage detection method based on light source vision lens recognition, which solves the problems of insufficient garbage classification accuracy and automation level of the existing garbage detection methods.

[0005] To achieve the above objectives, the present invention is realized through the following technical solutions: An intelligent garbage detection method based on light source vision lens recognition, including the following steps: S1. Capture garbage and environmental image data through a vision sensor; S2. Use a quantum optical enhanced vision lens system to collect the image data; S3. Perform tensor decomposition on the obtained image data, and use the CANDECOMP / PARAFAC decomposition method to convert multi-dimensional image data into a low-dimensional matrix to obtain a compressed data representation; S4. Input the transformed low-dimensional matrix into a deep neural network for training the classification model, first identify the garbage objects in the image, and then perform feature extraction and classification; S5. According to the environmental image data, calculate the robot path through a path optimization algorithm, enabling the robot to complete the garbage grasping task in a changing environment.

[0006] Preferably, the environmental image includes data of at least one dimension, selected from RGB images, depth images, or multi-dimensional image data; The quantum optical enhanced vision lens system includes a quantum superposition state light source and a quantum entanglement state light source, and forms an enhanced light field distribution during image acquisition by adjusting the phase and amplitude of the light source.

[0007] Preferably, the tensor decomposition is achieved by a high-order tensor decomposition method, using CANDECOMP / PARAFAC decomposition; The CANDECOMP / PARAFAC decomposition method extracts the main spatial features and texture information in the image through a step-by-step optimization method, and stores them in the form of a low-dimensional matrix to reduce data redundancy.

[0008] Preferably, the steps of the tensor decomposition include: Organize the image data into a high-dimensional tensor, where each dimension in the tensor represents different attributes or spatial features of the image; Apply CANDECOMP / PARAFAC decomposition to the high-dimensional tensor, and decompose the tensor into a series of low-dimensional matrices, where each matrix represents an important feature of the original data; The low-dimensional matrix is further processed by a deep neural network after decomposition to ensure that the garbage objects in the image are effectively identified and classified.

[0009] Preferably, the path optimization algorithm performs path optimization based on the variational method and the steepest descent method, and the steps of path optimization include: On the basis of considering the environmental map and obstacles, establish the kinematic model of the robot, optimize the path through the variational method, and calculate the optimal driving trajectory; Use the steepest descent method to perform real-time adjustment of the path, correct the path according to the changes in dynamic environmental data, and drive the robot to efficiently reach the target position for garbage pickup.

[0010] Preferably, the environmental data in the path optimization includes the current position of the robot, the positions of obstacles in the environment, and the expected target position; The path optimization updates the environmental data in real time, combines the variational method to optimize the cost function of the path, and combines the steepest descent method to correct the path, enabling the robot to perform the garbage pickup task in an unknown or dynamically changing environment.

[0011] Preferably, the classification model is trained using the quantum annealing method; The training process includes optimizing the objective function through the quantum annealing method; The quantum annealing method improves the classification performance of the garbage classification model by adjusting the parameters of the quantum state.

[0012] Preferably, the objective function in the quantum optimization algorithm includes a classification loss function and a regularization term; The objective of the objective function is to minimize the classification error, and it optimizes the accuracy of garbage classification by adjusting parameters through quantum optimization.

[0013] Preferably, the training process of the classification model is executed through a quantum computing platform; The quantum computing platform is configured with a quantum processing unit, which is responsible for quantum optimization of the model parameters; The quantum optimization process is realized through the quantum annealing method, mainly adjusting the weights and biases of the model.

[0014] Preferably, the acquisition of the image data is through a quantum optical enhanced vision lens system, which is adjusted according to the quantum characteristics of the light source, enhances the tiny details in the acquired image, and makes the garbage features in the image more obvious.

[0015] The present invention provides an intelligent garbage detection method based on light source vision lens recognition. It has the following beneficial effects: 1. Through the collaborative action of the quantum superposition state light source and the quantum entanglement state light source, the present invention realizes the enhancement processing of garbage image data, thereby improving the clarity and detail resolution of the image. Compared with the traditional optical imaging system, the present invention can effectively reduce the interference of environmental noise, make the details and edges of garbage objects more obvious, greatly improve the accuracy of the garbage classification system, provide clearer and richer input data for subsequent deep learning classification, and promote the improvement of the garbage classification accuracy.

[0016] 2. By using the high-order tensor decomposition technology to perform dimensionality reduction processing on the image data, the present invention can convert the image data into a low-dimensional matrix while retaining the key feature information. Compared with the traditional dimensionality reduction technology, the tensor decomposition method of the present invention effectively reduces the redundant data, simplifies the computational complexity of the subsequent deep neural network, and at the same time improves the accuracy of the garbage classification system, especially showing superiority in the recognition and classification of garbage objects in complex scenarios, providing a technical guarantee for realizing automatic and efficient garbage classification.

[0017] 3. In the training process of the garbage classification model, the present invention introduces a quantum annealing optimization algorithm, enabling it to accelerate the training process of the garbage classification model and optimize the classification accuracy of the model. Compared with the traditional optimization method based on gradient descent, quantum optimization can find the global optimal solution faster, reduce the training time, and at the same time improve the automated decision-making ability of the model in complex environments, further promoting the improvement of the automation level of the intelligent garbage classification system.

[0018] 4. In the robot path optimization, the present invention combines the variational method and the steepest descent method to achieve path planning and dynamic adjustment of the robot. Compared with the traditional path planning method, the path optimized by the variational method can dynamically adapt to the changing environment, and the real-time adjustment combined with the steepest descent method enables the robot to efficiently and accurately complete the garbage grasping task in complex environments, greatly improving the performance and efficiency of the robot in the automated garbage collection process and enhancing the overall automation level. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a flowchart of the method steps of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0020] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0021] Please refer to the attached Figure 1 , an intelligent garbage detection method based on light source vision lens recognition provided by an embodiment of the present invention includes the following steps: S1. Capture garbage and environmental image data through a vision sensor; Specifically, in this embodiment, step S1 is mainly used to obtain garbage image data, which is the environmental image captured by the vision sensor. The environmental image, as the input data, is crucial for the subsequent garbage recognition and classification processes. The main purpose of this step is to ensure that the image data has sufficient quality for effective garbage object detection in subsequent image processing and analysis.

[0022] Generally, a visual sensor uses a camera or other optical sensing devices, which may include infrared cameras, RGB cameras, depth cameras, etc., for acquiring different types of environmental images. Specifically, the image data that the visual sensor can capture should cover multi-dimensional information of the environment, such as spatial features, color features, and depth information of the image. These information are of great significance for the accuracy of garbage detection and classification.

[0023] In a possible implementation, the visual sensor can capture three-dimensional image data in the environment, including depth scanning of objects in the environment, and the obtained image data includes pixel point information and depth coordinate information of the image. The combination of such multi-dimensional information enables garbage detection to not only rely on the content of the two-dimensional image, but also further improve the recognition accuracy of garbage objects through depth information.

[0024] Furthermore, when acquiring image data, the visual sensor in this embodiment can perform signal enhancement in combination with certain preprocessing steps. Assume that the captured environmental image data is an image matrix I(x,y), where x and y respectively represent the pixel coordinates in the image, and I represents the brightness value or color value of the pixel point. The processing of the image data can be optimized using the following image enhancement formula: I ′ (x,y) = α·I(x,y) + β; where, I ′ (x,y) represents the enhanced image matrix; α is the contrast adjustment coefficient; β is the brightness adjustment constant. By appropriately adjusting the values of α and β, the details and contrast in the image can be enhanced, and the image quality can be improved for subsequent garbage recognition.

[0025] After the image data is acquired, in the subsequent processing steps, operations such as image denoising and edge enhancement may also be involved, and these steps also need to be optimized with the help of algorithms. For example, Gaussian filtering or median filtering can be used to remove noise points in the image to obtain smoother image data.

[0026] As an option, the visual sensor can capture continuous environmental image data in real time to form a dynamic image stream for subsequent real-time garbage detection. By continuously acquiring environmental images, it can ensure that in a dynamic environment, the positions and states of garbage objects are updated in a timely manner, thereby further enhancing the real-time performance and accuracy of garbage recognition.

[0027] In this embodiment, the visual sensor can work in cooperation with other sensor systems to form a multi-sensor fusion system. For example, sensor fusion technology can combine data from different sensors (such as temperature sensors, pressure sensors, etc.) to provide more comprehensive environmental information, thereby providing more accurate data support for subsequent garbage object detection.

[0028] Through the above technical means, the environmental image data obtained by the vision sensor in this embodiment can effectively support subsequent garbage recognition and classification processing, especially under complex or dynamic environmental conditions, ensuring the reliability and accuracy of garbage detection results.

[0029] S2. Use a quantum optical enhanced vision lens system to collect the image data; Specifically, in this embodiment, step S2 involves using a quantum optical enhanced vision lens system to collect and enhance the environmental image data obtained from step S1. The core purpose of this step is to utilize the quantum optical properties to optimize the captured image quality by adjusting the phase and amplitude of the light source, so as to ensure the accuracy of subsequent garbage detection and classification.

[0030] Generally, a quantum optical enhanced vision lens system includes a quantum superposition state light source and a quantum entanglement state light source. The quantum superposition state light source can generate multiple possible light states, while the quantum entanglement state light source generates a stronger optical interference effect by entangling multiple light waves. In this way, the light field output by the light source can enhance the coherence during image acquisition and improve the image detail performance. Specifically, by adjusting the phase and amplitude, the light source can effectively increase the coherence of the light wave, making the signal-to-noise ratio of image acquisition significantly improved.

[0031] For example, the process of collecting image data can be described by the following formula: I ′ (x, y) = η · I(x, y) · cos(φ); where, I ′ (x, y) represents the image data enhanced by the quantum optical system; I(x, y) is the original image data; η is the light source gain factor; φ is the light source phase difference; (x, y) are the image pixel coordinates. In this formula, cos(φ) describes the influence of the light source phase difference on the image enhancement effect. By precisely controlling the phase difference, the clarity of the image can be improved, especially in low-light or high-contrast scenarios, and the enhancement effect of the light field is particularly obvious.

[0032] In some embodiments, the light source of the quantum optical enhanced vision lens system adjusts the image light field by using the quantum interference effect. Through the phase interference between quantum light sources, the texture and edge details in the image are enhanced, especially the recognizability of small objects is improved. Through this technology, the system can always maintain clear and high-quality images under different ambient light conditions.

[0033] As an option, the quantum optical enhanced vision lens system can incorporate autofocus technology to adjust the focal length in real time according to the captured image information. This autofocus can ensure that the waste objects are accurately and clearly captured regardless of the depth. For example, in some cases, the image may contain objects at multiple levels. Through the autofocus technology, the system can dynamically focus on the target waste object, further improving the recognition accuracy.

[0034] In a possible implementation, the quantum optical system can be based on a laser as the light source. The laser has characteristics such as strong coherence and stable wavelength, and can generate highly consistent light waves. By adjusting the output beam frequency and phase of the laser, the light source can have a highly concentrated light field, thereby improving the resolution of the image. For example, by adjusting the beam intensity and phase difference of the light source, clear image data can be obtained in low-light environments, ensuring that the recognition of waste objects is not affected by the environment in different scenarios.

[0035] In addition, in some embodiments, the quantum optical enhanced vision lens system can further reduce image distortion caused by factors such as atmospheric interference or lens aberration through adaptive optical technology. By quickly adjusting the optical elements in the lens, the system can obtain more stable and accurate image data in a dynamic environment.

[0036] In summary, the quantum optical enhanced vision lens system in this embodiment not only improves the image quality by flexibly applying quantum optical effects and automatic adjustment mechanisms, but also ensures that the system can always obtain image data with high resolution and clarity under different environmental conditions, laying a solid foundation for subsequent waste recognition and classification.

[0037] S3. Perform tensor decomposition on the obtained image data, and use the CANDECOMP / PARAFAC decomposition method to convert the multi-dimensional image data into a low-dimensional matrix to obtain the compressed data representation.

[0038] Specifically, in this embodiment, step S3 involves performing tensor decomposition processing on the image data obtained from step S2. The purpose of this step is to convert the high-dimensional image data into a low-dimensional matrix representation, thereby simplifying the subsequent image feature extraction and classification processes. This process provides a simplified data representation that is convenient for calculation for the subsequent processing of the deep neural network.

[0039] In general, image data contains a large amount of pixel information. Directly processing this high-dimensional data consumes a large amount of computing resources and may lead to performance bottlenecks. Therefore, reducing the dimension of image data through tensor decomposition methods can effectively reduce data redundancy and extract key features in the image. Specifically, tensor decomposition techniques can transform multi-dimensional image data into a set of low-dimensional matrices through the decomposition of high-order tensors. These matrices can contain the main information of the image and have good representation capabilities.

[0040] In this embodiment, the CANDECOMP / PARAFAC decomposition method is used to perform tensor decomposition on the image data. This method represents the image data as a high-order tensor and decomposes it into multiple low-dimensional matrices. Each item of these low-dimensional matrices represents a part of the features of the image data. The goal of CANDECOMP / PARAFAC decomposition is to find a set of factor matrices such that their product can approximate the original high-order tensor.

[0041] Specifically, the image data is first organized into a three-dimensional or higher-dimensional tensor. Each dimension of the tensor represents an attribute or feature of the image, such as color, spatial coordinates, brightness, etc. Assume the tensor of the original image data is: where I, J, and K are the three dimensions of the tensor, representing different features of the image; represents the set of real numbers.

[0042] Then, by applying the CANDECOMP / PARAFAC decomposition method, the tensor is decomposed into several factor matrices: where R is the rank of the decomposition, representing the dimension of the low-dimensional matrices after decomposition. The goal of this decomposition process is to minimize the following objective function: where a r 、b r 、c r are the r-th columns in matrices A, B, and C respectively; represents the outer product operation of tensors; ∥.∥ F represents the Frobenius norm, which measures the error between the decomposition result and the original tensor.

[0043] In a possible implementation, after CANDECOMP / PARAFAC decomposition, the obtained low-dimensional matrices contain the main features in the image data, such as key information like texture and edges. After this process, the originally complex image data is simplified into a set of matrix representations with smaller dimensions, which is convenient for subsequent deep learning networks to perform feature extraction and classification.

[0044] As an option, the low-dimensional matrices obtained after tensor decomposition can be directly input into a deep neural network for further feature learning. In some embodiments, the structure of the deep neural network includes multiple convolutional layers, pooling layers, and fully connected layers, which can effectively extract high-level features of the image from the decomposed low-dimensional matrices for garbage classification and detection.

[0045] In a further implementation, to improve the computing efficiency, the low-dimensional matrix can be quantized to reduce the amount of computation and improve the processing speed. Specifically, in the representation of the low-dimensional matrix, some unimportant elements can be compressed by quantization, thus achieving efficient storage and processing of data.

[0046] In this embodiment, the dimensionality reduction of the image data by tensor decomposition can effectively reduce the complexity of the data while retaining the main features of the image, which provides a simplified and highly expressive data representation for subsequent garbage object detection. This technology can be effectively applied in various environments to improve the accuracy and processing speed of garbage detection.

[0047] S4. Input the transformed low-dimensional matrix into a deep neural network for training a classification model to first identify the garbage objects in the image and then perform feature extraction and classification; Specifically, in this embodiment, step S4 involves feature extraction from the image data after tensor decomposition in step S3. The core objective of this process is to obtain key features that can characterize the image content through specific mathematical transformations and statistical analyses. This step not only provides the basic data for subsequent garbage recognition and classification but also ensures the efficiency and robustness of the computation.

[0048] Generally, the method of feature extraction is selected according to the dimension and data distribution of the image data. Since the data has been mapped to a low-dimensional space after tensor decomposition, methods such as principal component analysis (PCA), independent component analysis (ICA), or non-negative matrix factorization (NMF) can be used to further extract features. Specifically, the goal is to extract the most discriminative features for subsequent training and inference of the classification model.

[0049] In this embodiment, first, the low-dimensional matrices A, B, and B after tensor decomposition are normalized to eliminate the influence of data scale differences on the feature extraction process. Assume the normalized form of matrix A is A ′ , and its calculation method is as follows: where μ A and σ A represent the mean and standard deviation of each element of matrix A respectively; A ′ij is the element at position (i, j) in the standardized matrix A. A similar normalization process also applies to matrices B and C.

[0050] As an option, key features can be further extracted based on singular value decomposition (SVD). The SVD method can further decompose a low-dimensional matrix to obtain the most principal eigenvectors. In some embodiments, matrix A ′ after SVD decomposition, can be expressed as: where U A and V A are the left singular matrix and the right singular matrix respectively; S A is a diagonal matrix, and its diagonal elements represent singular values; represents the transpose of matrix V A , that is, by interchanging the rows and columns of V A , we get These singular values represent the main information content of the data and can be used for feature selection.

[0051] Specifically, when performing feature selection, a threshold τ can be set, and only the feature components that satisfy the following condition are retained: s i ≥τ·max(S A ); where s i is the i-th singular value of S A ; τ is an empirically set weight parameter; max(S A ) is the maximum value in the set S A . For example, in some cases, selecting the feature components with the top 95% cumulative energy contribution can ensure data compression while maximizing the retention of information.

[0052] In a possible implementation, Gabor filters can also be used to perform spatial frequency analysis on image data to obtain local features such as edges and textures. Assume the response function of the Gabor filter is: where (x ′ , y ′ ) are the coordinates after rotation transformation; x and y represent the coordinates in the image; 2π is a constant representing a multiple of pi; θ is the filter direction; λ is the wavelength; σ controls the scale; γ is the spatial aspect ratio parameter; G(x, y, θ, λ) represents certain filters in image processing, especially in processing image textures, edge detection, etc. By adjusting these parameters, feature response values at different scales and directions can be obtained.

[0053] In some embodiments, to further improve the robustness of feature extraction, a feature extraction method based on deep learning, such as a convolutional neural network (CNN), can be combined. In this case, the data after tensor decomposition can be used as input, and deep features can be extracted through multiple convolutional layers and pooling layers.

[0054] Generally, the feature map F of the l-th layer of the convolutional neural network l can be expressed as: F l = f(W l * F l-1 + b l ); where W l is the convolutional kernel weight; b l is the bias term; f(.) is the non-linear activation function; * represents the convolution operation; F l-1 is the feature map of the previous layer.

[0055] Through the above method, the extracted features can be used as the input of the garbage recognition model. The dimension of the feature vector is related to the task requirements. Either fixed-length features can be used, or the attention mechanism can be combined for dynamic feature selection to improve the accuracy of garbage classification.

[0056] S5. According to the environmental image data captured by the vision sensor, the robot path is calculated through a path optimization algorithm. The path optimization algorithm uses the variational method to optimize the path and combines the steepest descent method for dynamic adjustment of the path, enabling the robot to complete the garbage grasping task in a changing environment; Specifically, in this embodiment, step S5 involves classifying and recognizing the image using the features extracted in the previous steps. The goal of this step is to classify the garbage image and determine its category by constructing a suitable classification model. The accuracy of classification and recognition directly affects the performance of the system. Therefore, the selection of the classification algorithm and parameter adjustment are crucial in this step.

[0057] Generally, the classification process will select a suitable algorithm based on the extracted features and task requirements. In some application scenarios, traditional machine learning methods such as support vector machine (SVM), k-nearest neighbor (k-NN), and decision tree can be used as candidate algorithms. On the other hand, in recent years, deep learning methods such as convolutional neural network (CNN) and long short-term memory network (LSTM) have been widely used in image classification tasks due to their strong feature learning ability.

[0058] In this embodiment, a convolutional neural network (CNN) is used for garbage classification. Specifically, at the beginning of step S5, the features extracted from step S4 are first input into the CNN network. In some embodiments, the feature map undergoes a convolution operation through a convolutional layer and undergoes downsampling through a pooling layer, thereby gradually extracting high-level abstract features. Assume that at the l-th layer, the result of the convolution operation is F l , which can be expressed as the l-th layer feature map F of the convolutional neural network in step S4 l formula.

[0059] As an option, after the convolution operation, the pooling layer reduces the dimension of the feature map, usually using max pooling MaxPool(F l ) or average pooling AveragePool(F l ). For example, assume that the feature map after pooling is F p , and its representation is: F p =MaxPool(F l ); In a possible implementation, the size and stride of the pooling operation are adjusted according to the task requirements, which directly affects the performance of the model.

[0060] Then, the pooled feature map is further processed through several fully connected layers, and the probability distribution of each garbage category is obtained through the Softmax activation function. Specifically, the output vector y i of the last fully connected layer can be expressed as: where z i is the i-th component of the output; C is the number of categories; exp(.) represents the exponential function; the exponential function is used to convert the output of the network into the probability distribution of each category.

[0061] In some embodiments, the classification model can further adopt transfer learning techniques to improve the adaptability and generalization ability of the model on different datasets. In this case, using a pre-trained CNN model, such as VGG16, ResNet, etc., for transfer learning can accelerate the training process and effectively reduce the overfitting phenomenon.

[0062] Generally, in order to improve the robustness of classification, some techniques may also be adopted, such as data augmentation (including operations such as rotation, translation, and scaling) and dropout regularization techniques. These techniques help to improve the generalization ability of the model during the training process and reduce the overfitting of the model to the training data.

[0063] In some embodiments, classification is performed by combining multimodal data. Specifically, in addition to image features, other information related to the image can be introduced, such as sensor data, environmental information, etc. These multimodal data are processed through joint learning, which can further improve the accuracy of classification.

[0064] After the classification process, a prediction result of the garbage category will be finally output. This prediction result will be fed back into the system for performing subsequent actions, such as garbage sorting, detection, or disposal operations.

[0065] The environmental image includes data of at least one dimension selected from RGB images, depth images, or multi-dimensional image data; the quantum optical enhanced vision lens system includes a quantum superposition state light source and a quantum entanglement state light source, and by adjusting the phase and amplitude of the light source, an enhanced light field distribution during the image acquisition process is formed; Tensor decomposition is achieved through a high-order tensor decomposition method, using CANDECOMP / PARAFAC decomposition; The CANDECOMP / PARAFAC decomposition method extracts the main spatial features and texture information in the image through a step-by-step optimization method and stores them in the form of a low-dimensional matrix to reduce data redundancy; The steps of tensor decomposition include: Organize the image data into a high-dimensional tensor, where each dimension in the tensor represents different attributes or spatial features of the image; Apply CANDECOMP / PARAFAC decomposition to the high-dimensional tensor, and decompose the tensor into a series of low-dimensional matrices, where each matrix represents an important feature of the original data; The low-dimensional matrices are further processed through a deep neural network after decomposition to ensure that the garbage objects in the image are effectively identified and classified; The path optimization algorithm performs path optimization based on the variational method and the steepest descent method. The steps of path optimization include: Based on considering the environmental map and obstacles, establish the kinematic model of the robot, optimize the path through the variational method, and calculate the optimal driving trajectory; Use the steepest descent method to perform real-time adjustment of the path, correct the path according to the changes in dynamic environmental data, and drive the robot to be able to efficiently reach the target position for garbage pickup; The environmental data in path optimization includes the current position of the robot, the positions of obstacles in the environment, and the expected target position; Path optimization updates the environmental data in real time, combines the variational method to optimize the cost function of the path, and combines the steepest descent method to correct the path, enabling the robot to perform garbage pickup tasks in an unknown or dynamically changing environment; The classification model is trained using the quantum annealing method; The training process includes optimizing the objective function through the quantum annealing method; The quantum annealing method improves the classification performance of the waste classification model by adjusting the parameters of the quantum state; The objective function in the quantum optimization algorithm includes a classification loss function and a regularization term; The objective of the objective function is to minimize the classification error, which adjusts the parameters through quantum optimization to optimize the accuracy of waste classification; The training process of the classification model is executed through a quantum computing platform; The quantum computing platform is configured with quantum processing units responsible for quantum optimization of the model parameters; The quantum optimization process is implemented through the quantum annealing method, mainly adjusting the weights and biases of the model; The acquisition of image data is through a quantum optical enhanced vision lens system, which adjusts according to the quantum characteristics of the light source, enhances the tiny details in the acquired image, and makes the waste features in the image more obvious.

[0066] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent garbage detection method based on light source visual lens recognition, characterized in that: The following steps are involved: S1, capture garbage and environment image data through visual sensors; S2. Collecting the image data using a quantum optical enhanced vision lens system; S3, performing tensor decomposition on the acquired image data, using the CANDECOMP / PARAFAC decomposition method to convert the multi-dimensional image data into a low-dimensional matrix to obtain a compressed data representation; S4, input the transformed low-dimensional matrix into the deep neural network to train the classification model, first identify the garbage objects in the image, and then perform feature extraction and classification; S5. Based on the environmental image data, the robot path is calculated through the path optimization algorithm, so that the robot can complete the garbage grabbing task in a changing environment.

2. The intelligent garbage detection method based on light source visual lens recognition according to claim 1 is characterized in that: The environment image includes data of at least one dimension selected from RGB image, depth image or multi-dimensional image data; The quantum optical enhanced vision lens system includes a quantum superposition state light source and a quantum entangled state light source, and forms an enhanced light field distribution during image acquisition by adjusting the phase and amplitude of the light source.

3. The intelligent garbage detection method based on light source visual lens recognition according to claim 1 is characterized in that: The tensor decomposition is achieved by a high-order tensor decomposition method, using CANDECOMP / PARAFAC decomposition; The CANDECOMP / PARAFAC decomposition method extracts the main spatial features and texture information in the image in a step-by-step optimization manner, and stores them in the form of a low-dimensional matrix to reduce data redundancy.

4. The intelligent garbage detection method based on light source visual lens recognition according to claim 1 is characterized in that: The steps of tensor decomposition include: Organize image data into high-dimensional tensors, where each dimension represents a different attribute or spatial feature of the image; Apply CANDECOMP / PARAFAC decomposition to the high-dimensional tensor to decompose the tensor into a series of low-dimensional matrices, each of which represents an important feature of the original data; The low-dimensional matrix is ​​further processed by a deep neural network after decomposition to ensure that garbage objects in the image are effectively identified and classified.

5. The intelligent garbage detection method based on light source visual lens recognition according to claim 1 is characterized in that: The path optimization algorithm performs path optimization based on the variational method and the steepest descent method. The steps of path optimization include: Based on the environmental map and obstacles, the robot's kinematic model is established, the path is optimized through the variational method, and the optimal driving trajectory is calculated; The path is adjusted in real time using the steepest descent method, and the path is corrected according to changes in dynamic environmental data, so that the robot can efficiently reach the target location for garbage pickup.

6. The intelligent garbage detection method based on light source visual lens recognition according to claim 5 is characterized in that: The environmental data in the path optimization includes the current position of the robot, the position of obstacles in the environment, and the expected target position; The path optimization updates environmental data in real time, optimizes the cost function of the path in combination with the variational method, and corrects the path in combination with the steepest descent method, so that the robot can perform garbage picking tasks in unknown or dynamically changing environments.

7. The intelligent garbage detection method based on light source visual lens recognition according to claim 1 is characterized in that: The classification model training adopts quantum annealing method; The training process includes optimizing the objective function by a quantum annealing method; The quantum annealing method improves the classification performance of the garbage classification model by adjusting the parameters of the quantum state.

8. The intelligent garbage detection method based on light source visual lens recognition according to claim 7 is characterized in that: The objective function in the quantum optimization algorithm includes a classification loss function and a regularization term; The objective of the objective function is to minimize the classification error, which adjusts parameters through quantum optimization to optimize the accuracy of garbage classification.

9. The intelligent garbage detection method based on light source visual lens recognition according to claim 1 is characterized in that: The training process of the classification model is performed through a quantum computing platform; The quantum computing platform is equipped with a quantum processing unit, which is responsible for quantum optimization of model parameters; The quantum optimization process is implemented through a quantum annealing method, which mainly adjusts the weights and biases of the model.

10. The intelligent garbage detection method based on light source visual lens recognition according to claim 1, characterized in that: The image data is collected through a quantum optical enhanced vision lens system, which is adjusted according to the quantum characteristics of the light source to enhance the tiny details in the collected image and make the garbage features in the image more obvious.

Images