A method, device and medium for adjusting the projection height of a HUD

Abstract

The present invention relates to the field of head-up display technology, and in particular, to a method, device and medium for adjusting the projection height of a HUD. The method comprises calculating the initial projection height of a head-up display HUD according to the initial eye height of the driver; during the HUD display process, if the driver shows negative emotions, calculating the latest projection height of the HUD according to the current eye height of the driver; and adjusting the projection height of the HUD according to the initial projection height and the latest projection height. The present invention recognizes that the height of the HUD needs to be adjusted according to the driver's emotions and changes in eye height, and adjusts the projection height of the HUD in real time based on changes in eye height. When the driver's eye height and accustomed observation angle are different, the projection height can also be adaptively adjusted to a suitable projection height.

Description

Technical Field

[0001] The present invention relates to the technical field of head-up display (HUD), and in particular to a method, device, and medium for adjusting the projection height of a HUD. Background Art

[0002] The application of Head-Up Display (HUD) technology in the automotive industry has become a trend, providing drivers with real-time information, which not only helps drivers improve their perception and reaction speed, but also further improves driving safety and convenience.

[0003] Due to different physiological characteristics such as eye height, and their habitual viewing angles, drivers often need to adjust the height of the HUD display. However, most current HUDs cannot be adjusted adaptively and can only be manually adjusted, which is particularly inconvenient and unsafe, especially while driving.

[0004] In view of this, the present invention is proposed. Summary of the Invention

[0005] The present invention aims to provide a method, device, and medium for adjusting the projection height of a HUD. These methods can identify when the HUD height needs to be adjusted based on the driver's mood and changes in eye height, and adjust the HUD projection height in real time based on changes in eye height. Furthermore, when the driver's eye height differs from their usual viewing angle, the projection height can be adaptively adjusted to an appropriate level.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] In a first aspect, the present invention provides a method for adjusting the projection height of a HUD, comprising:

[0008] Calculate the initial projection height of the head-up display (HUD) based on the driver's initial eye height;

[0009] During the HUD display process, the driver's emotions are recognized in real time based on the driver's facial image;

[0010] If the driver shows negative emotions, calculating the latest projection height of the HUD according to the driver's current eye height;

[0011] The projection height of the HUD is adjusted according to the initial projection height and the latest projection height.

[0012] In a second aspect, the present invention provides an electronic device, comprising:

[0013] at least one processor, and a memory communicatively coupled to the at least one processor;

[0014] The memory stores instructions that can be executed by at least one of the processors, and the instructions are executed by at least one of the processors so that the at least one processor can execute the above-mentioned HUD projection height adjustment method.

[0015] In a third aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, wherein the computer instructions are used to enable the computer to execute the above-mentioned HUD projection height adjustment method.

[0016] Compared with the prior art, the present invention has the following beneficial effects:

[0017] 1. During the HUD display process, the present invention identifies negative emotions in the driver based on the driver's facial image, and combines the projection height obtained by the human eye to identify negative emotions caused by inappropriate height, thereby capturing the driver's subjective feelings and adjustment needs for the HUD height. In this way, the HUD height can be adjusted based on the different needs of different drivers, rather than relying solely on human eye height.

[0018] 2. The present invention adjusts the HUD height based on the initial projection height and the latest projection height, that is, adjusts the HUD height according to changes in the human eye's line of sight. This dynamic adjustment method can ensure that the HUD projection content is within the driver's optimal line of sight in real time during the HUD display process.

[0019] 3. This invention uses a two-layer quantum tensor-enhanced convolutional model to identify driver emotions. This model leverages the long-range and short-range entanglement of quantum tensor networks to capture more image data and reduces model complexity due to its low number of model parameters. Furthermore, once quantum computers reach a certain level of development in the future, the supercomputing power of quantum computers can be leveraged to achieve millisecond-level processing speeds. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0021] Figure 1 This is a flow chart of a method for adjusting the projection height of a HUD provided by the present invention;

[0022] Figure 2 is a schematic diagram of the initial projection height provided by an embodiment of the present invention;

[0023] Figure 3Schematic diagram of the structure of the double-layer quantum tensor enhanced convolution model DbConTenNet provided by an embodiment of the present invention;

[0024] Figure 4 Schematic diagram of the internal structure of the feature extraction module and the feature fusion module provided in an embodiment of the present invention;

[0025] Figure 5 Schematic diagram of the internal structure of the data compression, flattening and feature mapping module, the MERA deep global feature extraction module, the MPS deep global feature extraction module and the classification module provided in an embodiment of the present invention;

[0026] Figure 6 Schematic diagram of data compression and flattening operations provided by an embodiment of the present invention;

[0027] Figure 7 is a schematic diagram of the execution process of the classification module provided by an embodiment of the present invention;

[0028] Figure 8 Schematic diagram of an MPS approximate representation of a weight tensor provided by an embodiment of the present invention;

[0029] Figure 9 It is a structural diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0030] The following description of exemplary embodiments of the present application is made in conjunction with the accompanying drawings, including various details of the embodiments of the present application to facilitate understanding. These details should be considered as merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present application. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0031] As mentioned in the background, existing HUDs lack adaptive adjustment capabilities and require manual adjustment, which can be particularly inconvenient and unsafe, especially while driving. While some automated HUD height adjustment solutions exist, these are often based on fixed values ​​for the driver's eye height and height, and fail to consider the driver's subjective experience, resulting in unsatisfactory adjustment results. The present invention will be further described in detail below with reference to examples.

[0032] Example 1

[0033] Figure 1This is a flow chart of a method for adjusting the projection height of a HUD provided by the present invention. This embodiment is applicable to adjusting the projection height of a HUD system on a vehicle. This method can be performed by a HUD projection height adjustment device, which can be composed of software and / or hardware and is generally integrated into an electronic device. The electronic device can be a vehicle-mounted computer or a device independent of the vehicle-mounted computer. The electronic device can control the projection direction of the HUD projection system to change the projection height.

[0034] like Figure 1 As shown, this embodiment provides a method for adjusting the projection height of a HUD, comprising the following steps:

[0035] S110 : Calculating an initial projection height of a heads-up display (HUD) according to the driver's initial eye height.

[0036] Optionally, after the seat pressure sensor detects that the driver has sat down, or after the driver has adjusted the seat height and angle, the driver's eye height is captured using an in-car camera or eye tracker. For ease of description and distinction, the captured eye height is referred to as the initial eye height, which is the driver's eye height when looking forward before the HUD is projected.

[0037] In actual applications, the projection height of the HUD should not block the driver's forward vision, nor overlap with the image of the front hood. The initial projection height should be between the initial eye height and the height of the front hood.

[0038] Optionally, based on the driver's initial eye height, the projection position of the eyes in the direction of the vertical horizontal plane is calculated; and the midpoint height between the projection position and the front hood position is used as the initial projection height Q of the HUD. Figure 2 This is a schematic diagram of the initial projection height provided by an embodiment of the present invention. The projection position of the eye in the direction perpendicular to the horizontal plane is point H. The intersection of the driver's eyes and the front hood edge is point h, i.e., the front hood position. Q is calculated using the following formula.

[0039] ; (1)

[0040] This embodiment uses the midpoint height between the projection position and the front hood position as the initial projection height, which neither blocks the driver's forward vision nor overlaps with the image of the front hood, providing a benchmark for subsequent changes in the latest projection height.

[0041] S120 . During the HUD display process, recognize the driver's emotions based on the driver's facial image.

[0042] Once the HUD begins projecting, it displays navigation information, weather information, vehicle status information, and more in front of the driver. The driver uses this projection to assist in driving decisions. If the HUD is projected at an inappropriate height, the driver will look up, crane their neck, or look down to see the content, expressing negative emotions such as nervousness and anxiety. However, if the HUD is projected at the appropriate height, the driver will express positive emotions such as calmness and joy.

[0043] In a specific implementation, the driver's facial image is captured by an in-car camera, and an emotion recognition model is used to identify the driver's emotions based on the facial image. The in-car camera needs to capture the driver's face and can be located above the front windshield or on the dashboard.

[0044] The emotion recognition model can be a deep neural network model that extracts features and classifies facial images to obtain emotion recognition results. The results can be classified into positive and negative emotions, or specific emotions such as anxiety, tension, calmness, appreciation, and joy. These specific emotions are classified as positive or negative.

[0045] S130: Determine whether the driver's emotion is positive or negative. If it is positive, execute S131; if it is negative, execute S132.

[0046] S131. End the operation.

[0047] If the driver is expressing positive emotions, such as calmness, appreciation, and joy, it indicates that the driver can accurately perceive the HUD display content at the current HUD projection height, and there is no need to adjust the HUD projection height. The process then returns to S120 to continue real-time recognition of the driver's emotions. This is because the driver's sitting posture, seat height, and other factors may change during driving, and the desired HUD projection height may also change, necessitating real-time adjustment.

[0048] S132. Calculate the latest projection height of the HUD according to the current eye height of the driver.

[0049] If the HUD height is not appropriate, the driver will look up, stretch their neck, or look down to see the HUD projection content so that their line of sight can see the HUD projection content clearly. Based on this, the driver's current eye height is captured, specifically using an in-car camera or eye tracker to capture the driver's eye height, and the latest HUD projection height P is recalculated using formula (1).

[0050] S140: Adjust the projection height of the HUD according to the initial projection height and the latest projection height.

[0051] Referring to the above description, the initial projection height is the projection height that adapts to the initial eye height of the driver when the HUD is not projected. This height is used as the reference height to judge the change of the driver's eye height after the HUD is displayed. If the latest projection height is greater than the initial projection height, it means that the driver raises their head / stretches their neck compared to the normal state, and then the projection direction of the HUD is lowered; conversely, if the latest projection height is less than the initial projection height, it means that the driver lowers their head compared to the normal state, and the projection direction of the HUD is raised.

[0052] In a specific embodiment, a quantization scheme for adjusting the projection height is provided. First, according to the initial projection height and the latest projection height, the deviation between the initial projection height and the latest projection height is calculated. The deviation here can be the difference between the initial projection height and the latest projection height, or the relative difference F, see formula (2):

[0053] ; (2)

[0054] Using the proportional difference as the deviation can more accurately obtain the change range of the projection height.

[0055] Then, according to the size relationship between the initial projection height and the latest projection height, the adjustment direction of the HUD projection is determined; according to the size of the deviation, the adjustment amplitude of the HUD projection is determined. Specifically, if the latest projection height is greater than the initial projection height, the projection direction of the HUD is lowered; if the latest projection height is less than the initial projection height, the projection direction of the HUD is raised; according to the size of the deviation, the threshold interval where the deviation is located is determined; according to the threshold interval, the adjustment amplitude of the HUD projection is determined; among them, the larger the threshold interval, the greater the adjustment amplitude. Multiple threshold intervals can be set to adjust the projection height of the HUD in multiple levels and improve the applicability of the HUD to different viewing heights.

[0056] During specific implementation, it is necessary to determine the rising or falling amplitude according to both the size relationship between the latest projection height and the initial projection height and the deviation value. Set a class I threshold of 5% and a class II threshold of 2%. These 2 thresholds can determine three threshold intervals. Specifically, when P > Q, if F > 5%, lower the height by 5% * |H - h|; if 5% > F > 2%, lower the height by 2% * |H - h|; if 2% > F > 0%, do not adjust. When P < Q, if F > 5%, raise the height by 5% * |H - h|; if 5% > F > 2%, raise the height by 2% * |H - h|; if 2% > F > 0%, do not adjust.

[0057] The aforementioned threshold range can be obtained through experimental calibration. During testing, users may experience negative emotions due to other discomforts, such as making phone calls or experiencing complex road conditions. These may cause deviations within 2%. Therefore, this embodiment does not adjust when 2% > F > 0% to improve fault tolerance.

[0058] The embodiments of the present invention have the following technical effects:

[0059] During the HUD display process, an embodiment of the present invention identifies negative emotions in the driver based on their facial image. Combined with the projection height obtained by the human eye, the embodiment identifies negative emotions caused by inappropriate height, thereby capturing the driver's subjective perception of and adjustment needs for the HUD height. This allows for adjustment based on the varying needs of different drivers for the HUD height, rather than solely relying on eye height. The embodiment of the present invention adjusts the HUD height based on the initial projection height and the latest projection height, that is, based on changes in the human eye's line of sight. This dynamic adjustment method ensures that the HUD projection content is within the driver's optimal line of sight in real time during the HUD display process.

[0060] The embodiment of the present invention can determine the adjustment direction of the HUD projection based on the size relationship between the initial projection height and the latest projection height; and determine the adjustment amplitude of the HUD projection based on the size of the deviation, so as to quickly obtain the most suitable projection height for the driver, improving height adjustment efficiency and user experience.

[0061] Example 2

[0062] This embodiment, building on the previous one, further refines the driver's emotion recognition based on facial images. Specifically, during the HUD display process, the driver's facial image is captured in real time via an in-vehicle camera. This facial image is then fed into the two-layer quantum tensor-enhanced convolutional model DbConTenNet to obtain the driver's emotion recognition results.

[0063] This embodiment uses a two-layer quantum tensor-enhanced convolutional model to identify driver emotions. This model leverages the long-range and short-range entanglement of quantum tensor networks to capture more image data and reduces model complexity due to its low number of model parameters. Furthermore, when quantum computers reach a certain level of development in the future, the supercomputing power of quantum computers can be leveraged to achieve millisecond-level processing speeds.

[0064] Figure 3This is a schematic diagram of the structure of the two-layer quantum tensor-enhanced convolutional model DbConTenNet provided by an embodiment of the present invention. The two-layer quantum tensor-enhanced convolutional model includes a preprocessing module, a feature extraction module, a feature fusion module, a data compression and flattening and feature mapping module, a two-dimensional multi-scale entanglement renormalization ansatz (MERA) deep global feature extraction module, a matrix product state (MPS) deep global feature extraction module, and a classification module. The driver's facial image is input into the preprocessing module. After being processed sequentially by the feature extraction module, feature fusion module, data compression and flattening and feature mapping module, a two-dimensional multi-scale entanglement renormalization ansatz (MERA) deep global feature extraction module, a matrix product state (MPS) deep global feature extraction module, and a classification module, the classification module outputs the emotion recognition result.

[0065] The preprocessing module is used to convert the color facial images into grayscale images, reducing the data dimension and simplifying the processing process. The facial images are adjusted to the same size and resolution for easy model processing. At the same time, in order to prevent overfitting, the dataset data is normalized to and changes the size of the image, which also reduces the complexity of model training.

[0066] Figure 4 Schematic diagram of the internal structure of the feature extraction module and feature fusion module provided by the embodiment of the present invention. Figure 4 The feature extraction module includes the base network BaseNet and the Histogram of OrientedGradients (HOG) feature extraction network; BaseNet includes at least two dense blocks DenseBlock, Figure 4 Two DenseBlocks are shown. After feature extraction and processing, the preprocessed image is input into the BaseNet and HOG feature extraction networks respectively to obtain two sets of image features.

[0067] Specifically, a densely connected network (DenseNet) model, previously trained to perform classification tasks on ImageNet, a large-scale visual database used for visual object recognition software research, was selected for transfer learning. Two shallow DenseBlocks were selected as the BaseNet, which was then used for image feature extraction. The specific process is as follows: Feature extraction and processing are performed on the preprocessed image (including feature image extraction through convolutional layers, followed by normalization, activation function layers (ReLU), and pooling layers). The features are then input into the DenseBlocks. To facilitate the connection between adjacent DenseBlocks, convolution and average pooling are used as transition layers between adjacent DenseBlocks, reducing the number and size of feature maps. Optionally, BaseNet has a total of 38 convolutional layers.

[0068] HOG feature extraction is also used. This is a commonly used image feature description method that constructs feature descriptors by counting the directional histogram of the gradient within each small area in the image. This method has good adaptability to lighting changes and shadows, making feature extraction more robust for facial images under different lighting conditions. At the same time, facial images may be subject to occlusions, such as wearing glasses or masks, but HOG can be tolerant to these occlusions because it focuses on the edge and texture information in the image rather than each specific pixel of the image.

[0069] The feature fusion module is used to combine two sets of image features (i.e., the features output by BaseNet and the features output by HOG) to generate fused features. For example, the texture features of an image can be combined with the color features to provide more comprehensive information.

[0070] Figure 5 This diagram illustrates the internal structure of the data compression, flattening, and feature mapping module, the MERA deep global feature extraction module, the MPS deep global feature extraction module, and the classification module, all provided in this embodiment. The data compression, flattening, and feature mapping module flattens the fused features into multiple one-dimensional vectors based on the left-right symmetry of the face.

[0071] On a human face, the left eye is symmetrical with the right eye, the left eyebrow is symmetrical with the right eyebrow, and the nose and mouth are also bilaterally symmetrical. When the HUD projection height is inappropriate, the driver may experience negative emotions that lead to asymmetrical facial expressions, such as raised eyebrows or pursed lips. Therefore, by compressing the vectors according to the bilaterally symmetrical positions of the face, we can capture the characteristics of asymmetrical expressions and thus identify the negative emotions caused by the driver's height inappropriateness. Figure 6 This is a schematic diagram of the data compression and flattening operation provided by an embodiment of the present invention, which evenly divides the fusion features into four areas: top, bottom, left, and right. Figure 6The four areas of different colors on the left. Then extract the features of one channel at the symmetrical position in the left and right symmetrical areas, and form a new vector with the extracted features. Figure 6 Pixels at different positions are numbered, 1 and 4 are symmetrical positions, 5 and 8 are symmetrical positions, and new vectors 1, 4, 5, and 8 are formed in the upper left corner area. Correspondingly, new vectors are formed in each of the four areas. Each new vector is then compressed and flattened into a one-dimensional vector, forming Figure 6 The rightmost row vectors. Therefore, see Figure 6 , the Squeeze operation converts the image from The image blocks are reshaped into The stack is then flattened to obtain the feature dimension The information of the four regions is connected and then compressed and flattened.

[0072] The data compression, flattening and feature mapping module uses formula (3) to perform feature mapping on the features in each vector to obtain the mapped features; the mapped features of each vector are calculated using formula (4) to obtain the tensor product .

[0073] ; (3)

[0074] In formula (3), Indicates the Features, Indicates the The dimension of the features is from 1 to , that is, each are mapped into one dimensional vector.

[0075] Each vector will get a tensor product of the local state vector after passing through formula (4) .

[0076] ; (4)

[0077] In formula (4), Represents all the features of the input, that is, the input vector, represents the direct product operation, Represents the tensor product of the Nth feature. The calculation formula is shown in the above formula (3). At the same time, formula (4) can also be understood as the direct product of multiple monomer wave functions in physics to form a multi-body wave function. Joint feature mapping Basically mapping each image to A vector in the dimensional feature space, and for the number of channels For an image, the local feature map is applied to each channel separately, and the resulting spatial dimension is .

[0078] Continue to see Figure 5 The MERA deep global feature extraction module includes multiple MPS layers, with mapping layers connected between adjacent MPS layers. In each MPS layer, multiple tensor products are collapsed and merged through corresponding disentanglers and MPS blocks. In each mapping layer, the output vectors of all MPS blocks are remapped back to the two-dimensional image space and normalized (for example, by BatchNorm1d). The features in each normalized vector are feature mapped according to formula (3) to obtain mapped features. The mapped features of each vector are tensor-calculated according to formula (4) to obtain tensor products. In this way, the tensor products are collapsed and merged through the MPS layer, then mapped back to the two-dimensional image space through the mapping layer and the tensor products are calculated. Then, the tensor products are repeated through the MPS layer and the mapping layer, and so on, for a set number of times (for example, 3 times). The global features output by the last MPS layer are input into the MPS deep global feature extraction module for further feature extraction to obtain the relationship between long-distance data.

[0079] The disentangler is mainly used to treat all correlations at the same length scale similarly. The MPS block is used to shrink vectors using a linear model in a high-dimensional feature space.

[0080] The MPS deep global feature extraction module is used to input global features into an MPS block to obtain features for classification. Because MERA primarily extracts long-range and short-range entanglement relationships, data feature extraction ultimately requires the entire MPS block. Therefore, the global features obtained from the MERA layer are input into the MPS deep global feature extraction module to obtain features for classification.

[0081] The classification module, specifically the Quantum Tensor Network (QTN) classification module, is used to normalize the features to be classified (for example, through BatchNorm1d normalization) and input them into the MPS block to obtain the classification results, namely positive emotions and negative emotions.

[0082] The decision rule for the classification task is pre-established by the MPS block, see formula (5):

[0083] ; (5)

[0084] In formula (5), , M1 represents the number of categories of the classification task. Is to input data The function value mapped to the specific category m is defined as shown in formula (6):

[0085] ;(6)

[0086] For a graphical representation of formula (6), see Figure 7 , Represents the weight tensor in the two-layer quantum tensor enhanced convolution model, and the output dimension is dimension. Represents the vector after feature mapping.

[0087] From formula (5), we can see that the classification module will eventually select the one with the largest result value as the classification result. However, in actual practice, it is found that when the number of images increases, the size increases, and the dimension of the feature map is very high, the weight tensor The growth will be exponential. In order to solve this problem, we can introduce the idea of ​​QTN and convert the weight tensor As a Order tensor, using MPS to represent the weight tensor , as shown in formula (7):

[0088] (7)

[0089] For a graphical representation of formula (7), see Figure 8 , where T and M both represent the tensor after contraction, and It is the connection line identifier of each tensor, and N is the number of dimensions. and Used as a subscript to distinguish different tensors T and M. Indicates the Nth connecting line l , Indicates the Nth connecting line r . Representation and connection lines and connecting wires The connected tensor T, Representation and connection lines , connecting wire and connecting wires The tensor M is connected. Representation and connection lines , connecting wire and connecting wires The tensor M is connected. Representation and connection lines and connecting wires The concatenated tensor T.

[0090] Depend on Figure 8It can be seen that using MPS to represent the original high-order tensor (N+1 dimension) as a contraction of multiple 3rd-order tensors M and two 2nd-order tensors T greatly reduces the number of parameters of the weight coefficients, thereby achieving the effect of reducing the complexity of the parameter coefficients.

[0091] Example 3

[0092] like Figure 9 As shown, this embodiment provides an electronic device, including:

[0093] at least one processor; and

[0094] a memory communicatively connected to at least one of the processors; wherein,

[0095] The memory stores instructions executable by the at least one processor. The instructions, when executed by the at least one processor, enable the at least one processor to perform the aforementioned method for adjusting the projection height of a HUD. The at least one processor in the electronic device is capable of performing the aforementioned method, thereby providing at least the same advantages as the aforementioned method.

[0096] Optionally, the electronic device also includes interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. The various components are connected to each other using different buses and can be installed on a common motherboard or installed in other ways as needed. The processor can process instructions executed in the electronic device, including instructions stored in or on the memory to display graphical information of a GUI (Graphical User Interface) on an external input / output device (such as a display device coupled to the interface). In other embodiments, if necessary, multiple processors can be used with multiple memories, and / or multiple buses can be used with multiple memories. Similarly, multiple electronic devices can be connected (for example, as a server array, a group of blade servers, or a multi-processor system), with each device providing part of the necessary operations. Figure 9 A processor 301 is taken as an example.

[0097] Memory 302, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules, such as the program instructions / modules corresponding to the HUD projection height adjustment method in the embodiments of the present invention. Processor 301 executes the software programs, instructions, and modules stored in memory 302 to execute various functional applications and data processing of the device, thereby implementing the aforementioned HUD projection height adjustment method.

[0098] The memory 302 may primarily include a program storage area and a data storage area. The program storage area may store an operating system and at least one application required for a function; the data storage area may store data created based on the use of the terminal, etc. Furthermore, the memory 302 may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some instances, the memory 302 may further include a memory remotely located relative to the processor 301, and these remote memories may be connected to the device via a network. Examples of the aforementioned networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0099] The electronic device may further include: an input device 303 and an output device 304. The processor 301, the memory 302, the input device 303 and the output device 304 may be connected via a bus or other means. Figure 9 The bus connection is taken as an example.

[0100] The input device 303 can receive input digital or character information, and the output device 304 can include a display device, an auxiliary lighting device (e.g., an LED), a tactile feedback device (e.g., a vibration motor), etc. The display device may include, but is not limited to, a liquid crystal display (LCD), a light-emitting diode (LED) display, and a plasma display. In some embodiments, the display device may be a touch screen.

[0101] Example 4

[0102] This embodiment provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to cause the computer to execute the above-described method. The computer instructions on the computer-readable storage medium are used to cause the computer to execute the above-described method, thereby having at least the same advantages as the above-described method.

[0103] The medium in the present invention may adopt any combination of one or more computer-readable media. The medium may be a computer-readable signal medium or a computer-readable storage medium. The medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination thereof. More specific examples of media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this document, a medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device or device.

[0104] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0105] The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wire, optical cable, RF (Radio Frequency), etc., or any suitable combination of the foregoing.

[0106] Computer program code for performing the operations of the present invention may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0107] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this application can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this application can be achieved. This is not a limitation herein.

[0108] The above specific embodiments do not constitute a limitation on the scope of protection of this application. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application shall be included within the scope of protection of this application.

Claims

1. A method for adjusting the projection height of a HUD, characterized in that: include: Calculate the initial projection height of the head-up display (HUD) based on the driver's initial eye height; During the HUD display process, the driver's emotions are recognized in real time based on the driver's facial image; If the driver shows negative emotions, calculating the latest projection height of the HUD according to the current eye height of the driver; Adjusting the projection height of the HUD according to the initial projection height and the latest projection height; Among them, adjusting the projection height of the HUD according to the initial projection height and the latest projection height includes: calculating the deviation between the initial projection height and the latest projection height according to the initial projection height and the latest projection height; determining the adjustment direction of the HUD projection according to the size relationship between the initial projection height and the latest projection height; and determining the adjustment amplitude of the HUD projection according to the size of the deviation.

2. The method according to claim 1, characterized in that Calculating the initial projection height of the head-up display HUD according to the initial eye height of the driver includes: Calculating the projection position of the eyes in a direction perpendicular to the horizontal plane according to the initial eye height of the driver; The midpoint height between the projection position and the front hood position is used as the initial projection height of the HUD.

3. The method according to claim 1, characterized in that According to the magnitude relationship between the initial projection height and the latest projection height, determining the adjustment direction of the HUD projection includes: If the latest projection height is greater than the initial projection height, the projection direction of the HUD is lowered; If the latest projection height is less than the initial projection height, the projection direction of the HUD is raised; Determining the adjustment amplitude of the HUD projection according to the size of the deviation includes: Determine the threshold interval of the deviation according to the size of the deviation; Determining an adjustment amplitude of the HUD projection according to the threshold interval; The larger the threshold interval is, the larger the adjustment range is.

4. The method according to claim 1, characterized in that: During the HUD display process, the driver's emotions are recognized in real time based on the driver's facial image, including: During the HUD display process, the driver's facial image is collected in real time through the in-car camera; The facial image is input into a double-layer quantum tensor enhanced convolution model to obtain the recognition result of the driver's emotion.

5. The method according to claim 4, characterized in that The double-layer quantum tensor enhanced convolution model includes a preprocessing module, a feature extraction module, a feature fusion module, a data compression flattening and feature mapping module, a two-dimensional multi-scale entanglement renormalization MERA deep global feature extraction module, a matrix product state MPS deep global feature extraction module and a classification module, which are connected in sequence.

6. The method according to claim 5, characterized in that The feature extraction module includes a base network BaseNet and a HOG feature extraction network; the BaseNet includes at least two dense blocks DenseBlock; The feature extraction module is used to extract and process the preprocessed image and input it into the BaseNet and HOG feature extraction networks to obtain two sets of image features; The feature fusion module is used to splice the two sets of image features to obtain fused features; The data compression flattening and feature mapping module is used to compress the fused features into multiple one-dimensional vectors according to the left-right symmetrical positions of the face; perform feature mapping on the features in each vector to obtain mapped features; perform tensor calculation on the mapped features of each vector to obtain a tensor product; The two-dimensional multi-scale entangled renormalization proposed MERA deep global feature extraction module includes multiple MPS layers, and a mapping layer is connected between adjacent MPS layers; wherein, in each MPS layer, multiple tensor products are contracted and merged with the MPS block through the corresponding disentangler; in each mapping layer, the output vectors of all MPS blocks are remapped back to the two-dimensional image space and normalized, and the features in each vector after normalization are feature mapped to obtain mapped features, and the mapped features of each vector are tensor calculated to obtain tensor products; the global features output by the last MPS layer are input into the MPS deep global feature extraction module; The MPS deep global feature extraction module is used to input the global features into an MPS block to obtain features to be classified; The classification module is used to normalize the features to be classified and input them into the MPS block to obtain the classification results.

7. The method according to claim 6, characterized in that The step of compressing the fusion features into a plurality of one-dimensional vectors according to the left-right symmetrical positions of the face includes: The fused features are evenly divided into four regions: up, down, left, and right; Extract features at symmetrical positions in a bilaterally symmetrical region, and construct a new vector using the extracted features; Each of the new vectors is compressed into a one-dimensional vector.

8. An electronic device, characterized in that: include: at least one processor, and a memory communicatively coupled to at least one of the processors; The memory stores instructions executable by at least one of the processors, and the instructions are executed by at least one of the processors so that the at least one processor can execute the method according to any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that: The medium stores computer instructions, and the computer instructions are used to enable a computer to execute the method according to any one of claims 1 to 7.

Images