Urban traffic carbon emission measuring and calculating method and related device

Abstract

The invention discloses an urban traffic carbon emission calculation method and a related device, and relates to the technical field of carbon emission calculation, and the method comprises the steps: obtaining a remote sensing image of each road region of a target city; for each road area, based on the remote sensing image of the road area, detecting vehicles and identifying the detected vehicle type and power type of each vehicle; counting the number of vehicles in each vehicle type-power type combination in the road area according to the recognition result, and obtaining an emission factor matched with each vehicle type-power type combination; according to the number of vehicles of each vehicle type-power type combination in the road area and the matched emission factors, the vehicle carbon emission of each vehicle type-power type combination in the road area is determined; and determining the traffic carbon emission of the target city according to the vehicle carbon emission of various vehicle type-power type combinations of each road area of the target city. According to the urban traffic carbon emission calculation method disclosed by the invention, accurate calculation of the carbon emission can be realized.

Description

Technical Field

[0001] This application relates to the field of carbon emission measurement technology, and in particular to a method and related apparatus for measuring carbon emissions from urban transportation. Background Technology

[0002] As the global climate governance process continues to deepen, government regulatory agencies and industry organizations are placing increasingly stringent demands on the accuracy, timeliness, and compliance of corporate carbon emission accounting. Urban transportation systems, as a major source of carbon emissions, require accurate measurement of their emissions as the foundation for effective emission reduction and scientific regulation. How to accurately measure the carbon emissions of urban transportation systems is a pressing issue that needs to be addressed. Summary of the Invention

[0003] In view of this, this application provides a method and related apparatus for calculating urban traffic carbon emissions, which can accurately calculate urban traffic carbon emissions. The technical solution is as follows:

[0004] The first aspect of this application provides a method for measuring carbon emissions from urban transportation, including: acquiring remote sensing images of each road area in a target city;

[0005] For each road area: Based on remote sensing images of the road area, vehicles are detected and the vehicle type and power type of each detected vehicle are identified; based on the identification results, the number of vehicles of each vehicle type-power type combination in the road area is counted, and the emission factor matching each vehicle type-power type combination is obtained; based on the number of vehicles of each vehicle type-power type combination in the road area and the emission factor matching each vehicle type-power type combination, the carbon emissions of each vehicle type-power type combination in the road area are determined;

[0006] The traffic carbon emissions of the target city are determined based on the carbon emissions of vehicles of different vehicle type and powertrain combinations in each road area of ​​the target city.

[0007] In one possible implementation, determining the carbon emissions of each vehicle type-power type combination within the road area based on the number of vehicles in each vehicle type-power type combination within the road area and the emission factor matched to each vehicle type-power type combination includes:

[0008] From a pre-constructed urban vehicle driving sampling database for the target city, obtain the time-scale average mileage and energy consumption per unit mileage for each vehicle type-power type combination within the road area;

[0009] The carbon emissions of each vehicle type-power type combination in the road area are calculated based on the number of vehicles in each vehicle type-power type combination within the road area and the emission factor matched with each vehicle type-power type combination, combined with the time-scale average mileage and energy consumption per unit mileage of each vehicle type-power type combination in the road area.

[0010] In one possible implementation, obtaining the emission factor matching each of the vehicle model-powertrain type combinations includes:

[0011] From a pre-constructed emission factor lookup table, the emission factor corresponding to each of the vehicle model-power type combinations is obtained, and the emission factor matching each of the vehicle model-power type combinations is obtained. The emission factor lookup table contains multiple vehicle model-power type combinations and emission factors corresponding to each of the multiple vehicle model-power type combinations.

[0012] In one possible implementation, detecting vehicles and identifying the vehicle type and powertrain type of each detected vehicle based on the remote sensing image of the road area includes:

[0013] Vehicle detection is performed on the remote sensing image of the road area to obtain multiple vehicle location bounding boxes;

[0014] Based on the location bounding box of each vehicle, vehicle images are extracted from the remote sensing images of the road area to obtain the remote sensing image of each vehicle in the road area;

[0015] Based on the vehicle type and power type recognition model, the remote sensing images of each vehicle in the road area are used to identify the vehicle type and power type. The vehicle type and power type recognition model is trained in advance using remote sensing images of training vehicles labeled with vehicle type and power type.

[0016] In one possible implementation, the remote sensing image of the road area is a multimodal remote sensing image, which includes a registered high-resolution optical remote sensing image and a thermal infrared remote sensing image.

[0017] The remote sensing image of the road area is used to detect vehicles, resulting in multiple vehicle location bounding boxes, including:

[0018] Vehicle detection is performed on high-resolution optical remote sensing images in the multimodal remote sensing images of the road area to obtain multiple vehicle location bounding boxes;

[0019] The step of extracting vehicle images from the remote sensing image of the road area based on the location bounding box of each vehicle to obtain the remote sensing image of each vehicle in the road area includes:

[0020] Based on the location bounding box of each vehicle, vehicle images are extracted from the multimodal remote sensing images of the road area to obtain a multimodal remote sensing image of each vehicle in the road area.

[0021] In one possible implementation, the remote sensing image of each vehicle is a multimodal remote sensing image, which includes a registered high-resolution optical remote sensing image and a thermal infrared remote sensing image.

[0022] The vehicle type and powertrain type recognition model identifies the vehicle type and powertrain type of each vehicle in the remote sensing image of the road area, including:

[0023] For each vehicle in the road area, the optical image feature extraction module based on the vehicle type and power type recognition model extracts visual discrimination features that characterize the appearance structure and component details from the high-resolution optical remote sensing images of the vehicle in the multimodal remote sensing images.

[0024] Based on the vehicle model and power type identification model, the thermal infrared image feature extraction module extracts thermal infrared identification features that characterize the temperature distribution and spatial layout of heat sources from the multimodal remote sensing images of the vehicle.

[0025] The cross-modal feature fusion module based on the vehicle model and power type recognition model fuses the visual discrimination features with the thermal infrared identification features to obtain cross-modal fusion features;

[0026] The multi-task classification module based on the vehicle model and powertrain type recognition model predicts the vehicle model and powertrain type based on the cross-modal fusion features.

[0027] In one possible implementation, fusing the visual discrimination feature with the thermal infrared identification feature to obtain a cross-modal fusion feature includes:

[0028] An enhanced thermal infrared identification feature is generated by using the visual discrimination feature as the query and the thermal infrared identification feature as the key and value for attention calculation.

[0029] An enhanced visual discriminative feature is generated by using the original thermal infrared identification feature as the query and the visual discriminative feature as the key and value for attention calculation.

[0030] The enhanced thermal infrared identification features are fused with the enhanced visual discrimination features to obtain cross-modal fusion features.

[0031] In one possible implementation, the remote sensing image of the road area is a multimodal remote sensing image, which includes a high-resolution optical remote sensing image and / or a thermal infrared remote sensing image, as well as a hyperspectral optical remote sensing image.

[0032] The remote sensing image of the road area is used to detect vehicles and identify the vehicle type and powertrain of each detected vehicle, including:

[0033] Based on high-resolution optical remote sensing images and / or thermal infrared remote sensing images in the multimodal remote sensing images of the road area, vehicles are detected and the model and power type of each detected vehicle are identified.

[0034] By performing spectral analysis on the atmosphere above the road in the hyperspectral optical remote sensing image of the multimodal remote sensing image of the road area, the concentration distribution information of exhaust characteristic gases above the road area is generated.

[0035] Based on the power type identification result of each vehicle and in conjunction with the concentration distribution information, the final power type of each vehicle is determined.

[0036] In one possible implementation, determining the final power type of each vehicle based on the power type identification result of each vehicle, combined with the concentration distribution information, includes:

[0037] By spatially registering the concentration distribution information with the location of each vehicle, the local gas concentration at the location of each vehicle is determined;

[0038] The final power type of each vehicle is determined based on the power type identification result of each vehicle and the local gas concentration at the location of each vehicle.

[0039] In one possible implementation, the method for calculating urban transportation carbon emissions further includes:

[0040] If traffic carbon emission benchmark data that matches the target city in time and space is obtained, the traffic carbon emissions calculated for the target city will be compared with the traffic carbon emission benchmark data.

[0041] The emission factors in the emission factor comparison table are adjusted based on the comparison results so that the adjusted emission factor comparison table can be used for the subsequent calculation of the target city's transportation carbon emissions.

[0042] A second aspect of this application provides an electronic device, including at least one processor and a memory connected to the processor, wherein:

[0043] The memory is used to store computer programs;

[0044] The processor is used to execute the computer program so that the electronic device can implement any of the above-mentioned methods for calculating urban traffic carbon emissions.

[0045] A third aspect of this application provides a computer storage medium carrying one or more computer programs, which, when executed by an electronic device, enable the electronic device to implement any of the above-described methods for calculating urban traffic carbon emissions.

[0046] Using the above technical solution, the urban traffic carbon emission calculation method provided in this application first acquires remote sensing images of each road area in the target city. Then, for each road area, vehicles are detected based on the remote sensing images, and each detected vehicle undergoes dual identification of vehicle type and power type. Based on this identification result, the number of vehicles with each vehicle type-power type combination in the road area is counted, and the emission factor matching each vehicle type-power type combination is obtained. Based on the number of vehicles with each vehicle type-power type combination in the road area and the emission factor matching each vehicle type-power type combination, the carbon emission of each vehicle type-power type combination in the road area is determined. Finally, based on the carbon emission of various vehicle type-power type vehicles in each road area of ​​the target city, the city-level traffic carbon emission is determined. The urban traffic carbon emission calculation method provided in this application can identify the specific vehicle type and power type of each vehicle in the road area based on the remote sensing images of the road area. On this basis, the number of vehicles can be counted by vehicle type-power type combination, and then the carbon emission can be accurately calculated by vehicle type-power type combination. This application provides a method for calculating urban traffic carbon emissions based on remote sensing images. This method realizes automatic detection of vehicles in road areas and automatic identification of vehicle type and power type based on remote sensing images of road areas, and calculates refined carbon emissions accordingly. This method provides real-time, reliable and low-cost key data decision support for precise emission reduction policies, intelligent low-carbon management and systematic green transformation of urban traffic, and has significant technological advancement, economic feasibility and social application value. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0048] Figure 1 A flowchart illustrating the method for calculating urban traffic carbon emissions provided in this application embodiment;

[0049] Figure 2 This is a schematic diagram illustrating the process of detecting vehicles and identifying the vehicle type and power type of each detected vehicle based on a remote sensing image of a road area, as provided in an embodiment of this application.

[0050] Figure 3 This is a flowchart illustrating an implementation method for detecting vehicles and identifying the vehicle type and power type of each detected vehicle based on a multimodal remote sensing image of a road area, as provided in an embodiment of this application.

[0051] Figure 4 A schematic diagram of the vehicle model and powertrain identification model provided in this application embodiment;

[0052] Figure 5 This is a flowchart illustrating another implementation method for detecting vehicles and identifying the vehicle type and power type of each detected vehicle based on a multimodal remote sensing image of a road area, as provided in the embodiments of this application.

[0053] Figure 6 This is a schematic diagram of the urban traffic carbon emission measurement device provided in an embodiment of this application. Detailed Implementation

[0054] The embodiments of this application are described below with reference to the accompanying drawings. The terminology used in the implementation section of this application is for explaining specific embodiments only and is not intended to limit the scope of this application.

[0055] The embodiments of this application will now be described with reference to the accompanying drawings. Those skilled in the art will recognize that, with technological advancements and the emergence of new scenarios, the technical solutions provided in the embodiments of this application are equally applicable to similar technical problems.

[0056] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such terms are interchangeable where appropriate; this is merely a way of distinguishing objects with the same attributes in the embodiments of this application. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, so that a process, method, system, product, or apparatus that comprises a series of elements is not necessarily limited to those elements, but may include other elements not explicitly listed or inherent to those processes, methods, products, or apparatuses.

[0057] In the process of realizing this case, the inventors discovered that the current urban traffic carbon emission calculation scheme is to use a traffic monitoring system based on ground cameras to count the traffic flow in road areas and obtain the average emission factor. Based on the traffic flow and average emission factor in the road areas, the carbon emissions of vehicles in the road areas are determined, and then the overall traffic carbon emissions of the city are determined based on the carbon emissions of vehicles in each road area of ​​the city.

[0058] While the aforementioned methods can determine urban traffic carbon emissions, their calculation accuracy is not high. The inventors of this case, through research, discovered that the low accuracy of these methods stems primarily from their failure to consider the differences in carbon emissions among different vehicle types and powertrains. Based on this finding, the inventors continued their research and ultimately developed a method for calculating urban traffic carbon emissions that can more accurately measure urban traffic carbon emissions.

[0059] The method for calculating urban transportation carbon emissions provided in this application will be described below through the following embodiments.

[0060] Please see Figure 1 The diagram illustrates a flowchart of a method for calculating urban transportation carbon emissions according to an embodiment of this application. This method may include:

[0061] Step S101: Obtain remote sensing images of each road area in the target city.

[0062] It can acquire remote sensing images of a target city and obtain a road vector map of the target city (a digital map stored and expressed in vector data format that describes the spatial location and topological relationship of the road network of the target city). Based on the road vector map of the target city, the urban road network is divided into multiple independent road areas, and the remote sensing image of each road area is extracted from the remote sensing images acquired for the target city.

[0063] When extracting remote sensing images of each road area from remote sensing images acquired for a target city, the remote sensing images acquired for the target city can be preprocessed first, such as radiometric correction (eliminating atmospheric interference), geometric correction (registering map coordinates), orthorectification (correcting distortion caused by terrain undulations), etc., and then the remote sensing images of each road area can be extracted from the preprocessed remote sensing images.

[0064] Optionally, the remote sensing images in this embodiment may include, but are not limited to, some or all of the following types of remote sensing images: optical remote sensing images, thermal infrared remote sensing images, microwave remote sensing images, etc. Optionally, optical remote sensing images may include some or all of the following types of remote sensing images: high-resolution optical remote sensing images, hyperspectral optical remote sensing images.

[0065] Optionally, optical remote sensing images can be acquired based on satellites (such as WorldView-3 / 4, Gaofen-2, etc.), thermal infrared remote sensing images can be acquired based on UAVs equipped with infrared sensors, and microwave remote sensing images can be acquired based on synthetic aperture radar.

[0066] Step S102: For each road area, based on the remote sensing image of that road area, detect vehicles and identify the vehicle type and power type of each detected vehicle.

[0067] Specifically, for each road area, vehicle detection is first performed on the remote sensing image of the road area to obtain multiple vehicle location bounding boxes. Then, vehicle images are extracted from the remote sensing image of the road area based on each vehicle location bounding box to obtain the remote sensing image of each vehicle in the road area. Finally, based on the remote sensing image of each vehicle in the road area, the vehicle type and power type of each vehicle in the road area are identified.

[0068] Vehicle model recognition identifies which of several vehicle types (such as sedan, SUV, truck, bus, lorry, motorcycle, etc.) a vehicle belongs to, and vehicle power type recognition identifies which of several power types (such as gasoline vehicle, new energy vehicle, etc.) a vehicle belongs to.

[0069] Step S103: Based on the vehicle type and power type identification results for each vehicle, count the number of vehicles of each type-power type combination in the road area, and obtain the emission factor that matches each type-power type combination in the road area.

[0070] After obtaining the vehicle type and power type identification results for each vehicle in the road area, the number of vehicles with each type and power type combination in the road area is counted. For example, the number of gasoline sedans, new energy sedans, gasoline SUVs, and new energy SUVs are counted.

[0071] In determining carbon emissions, in addition to obtaining the number of vehicles for each vehicle type-power type combination in the road area, it is also necessary to obtain the emission factor that matches each vehicle type-power type combination in the road area.

[0072] In one possible implementation, a pre-built emission factor lookup table can be obtained. This table contains various vehicle-powertrain combinations and their corresponding emission factors. Then, the emission factors for each vehicle-powertrain combination in the road area can be retrieved from the lookup table. For example, the emission factor for gasoline-powered cars (the mass of carbon dioxide emitted per liter of gasoline combustion) and the emission factor for new energy vehicles (e.g., the mass of carbon dioxide produced per kilowatt-hour of electricity supplied by the grid) can be obtained, thus yielding the emission factors matching each vehicle-powertrain combination in the road area. An example of an emission factor lookup table is shown below:

[0073] Table 1. Example of emission factor comparison

[0074]

[0075] Step S104: Determine the carbon emissions of each vehicle type-power type combination in the road area based on the number of vehicles of each vehicle type-power type combination in the road area and the emission factor matched with each vehicle type-power type combination in the road area.

[0076] Vehicle carbon emissions can be calculated based on a combination of vehicle type and powertrain type. For example, the carbon emissions of new energy passenger cars can be determined based on the number of new energy passenger cars and the emission factors that match them, while the carbon emissions of gasoline passenger cars can be determined based on the number of gasoline passenger cars and the emission factors that match them.

[0077] Step S105: Determine the traffic carbon emissions of the target city based on the carbon emissions of various vehicle type-power type combinations in each road area of ​​the target city.

[0078] After obtaining the carbon emissions of various vehicle type-power type combinations in different road areas of the target city, the carbon emissions of various vehicle type-power type combinations in different road areas can be summarized to obtain the city-level traffic carbon emissions.

[0079] In one possible implementation, the carbon emissions of vehicles with the same vehicle type and powertrain combination in each road area of ​​the target city can be aggregated to obtain the total carbon emissions of each vehicle type and powertrain combination in the target city.

[0080] In another possible implementation, the carbon emissions of various vehicle type-power combination combinations in different road areas of the target city can be aggregated to obtain the total traffic carbon emissions of the target city.

[0081] The urban traffic carbon emission calculation method provided in this application can identify the specific vehicle type and power type of each vehicle in a road area based on remote sensing images. Based on this, the number of vehicles can be counted according to vehicle type-power type combinations, and the emission factor matching each vehicle type-power type combination can be obtained. Therefore, carbon emissions can be accurately calculated based on vehicle type-power type combinations. This application provides a precise urban traffic carbon emission calculation method based on remote sensing images. This method achieves automatic vehicle detection and automatic identification of vehicle type and power type in a road area based on remote sensing images, and calculates refined carbon emissions accordingly. This method provides real-time, reliable, and low-cost key data decision support for precise emission reduction policies, intelligent low-carbon management, and systematic green transformation of urban traffic, and has significant technological advancement, economic feasibility, and social application value.

[0082] In some embodiments of this application, a specific implementation process is described for detecting vehicles and identifying the model and power type of each detected vehicle based on remote sensing images of a road area.

[0083] In one possible implementation, such as Figure 2 As shown, the specific implementation process of detecting vehicles and identifying the model and power type of each detected vehicle based on remote sensing images of a road area can include:

[0084] Step S201: Perform vehicle detection on the remote sensing image of the road area to obtain multiple vehicle location bounding boxes.

[0085] In one possible implementation, vehicle detection can be performed on the remote sensing image of the road area based on a pre-trained vehicle detection model (such as YOLO or Faster R-CNN). The vehicle detection model can be trained using remote sensing images of the road area with vehicle locations marked.

[0086] Step S202: Extract vehicle images from the remote sensing images of the road area based on the location bounding box of each vehicle, and obtain the remote sensing image of each vehicle in the road area.

[0087] Based on the vehicle location bounding box, the image of each vehicle can be cropped from the remote sensing image of the road area to obtain the remote sensing image of each vehicle in the road area.

[0088] Step S203: Based on the vehicle type and power type recognition model, identify the vehicle type and power type of each vehicle in the remote sensing image.

[0089] In this embodiment, the vehicle model and power type recognition model is trained in advance using remote sensing images of training vehicles labeled with vehicle model and power type. During training, the training objective is to make the vehicle model and power type recognition model predict the same vehicle model and power type as labeled in the remote sensing images of the training vehicles.

[0090] Vehicle remote sensing images can be collected from real-world scenarios to construct a training dataset for vehicle model and power type recognition models. To improve the generalization ability and robustness of the vehicle model and power type recognition models, generative adversarial networks can be used to synthesize vehicle remote sensing images under different resolutions, lighting conditions, and weather conditions. These synthesized vehicle remote sensing images can then be added to the training dataset to enhance the diversity of the training dataset.

[0091] The remote sensing image of each vehicle extracted from the remote sensing image of the road area is input into the vehicle type and power type recognition model to identify the vehicle type and power type, and the identification results of the vehicle type and power type of each vehicle in the road area are obtained.

[0092] This embodiment is not limited to identifying the vehicle type and powertrain type of each vehicle's remote sensing image based on a vehicle type and powertrain type recognition model. In another possible implementation, vehicle type recognition can be performed on the remote sensing image of each vehicle based on a vehicle type recognition model, and powertrain type recognition can be performed on the remote sensing image of each vehicle based on a powertrain type recognition model. Specifically, the vehicle type recognition model is trained using remote sensing images of training vehicles labeled with vehicle type information, and the powertrain type recognition model is trained using remote sensing images of training vehicles labeled with powertrain type information.

[0093] In one possible implementation, the remote sensing image of each road area can be a single-modal remote sensing image, such as an optical remote sensing image, a thermal infrared remote sensing image, or a microwave remote sensing image. In order to obtain better recognition results, in another possible implementation, the remote sensing image of each road area can be a multimodal remote sensing image. The following describes the specific implementation process of detecting vehicles and identifying the vehicle type and power type of each detected vehicle based on the multimodal remote sensing image of a road area.

[0094] There are several ways to detect vehicles and identify the model and power type of each detected vehicle based on multimodal remote sensing images of a road area. Please refer to [link / reference]. Figure 3 The diagram illustrates a flowchart of one implementation method, which may include:

[0095] Step S301: Perform vehicle detection on the high-resolution optical remote sensing image in the multimodal remote sensing image of the road area to obtain multiple vehicle location bounding boxes.

[0096] In this embodiment, the multimodal remote sensing images of the road area include registered high-resolution optical remote sensing images and thermal infrared remote sensing images.

[0097] Considering the high spatial resolution and rich texture, color, and shape features of high-resolution optical remote sensing images, this embodiment uses high-resolution optical remote sensing images for vehicle detection.

[0098] Step S302: Based on the location bounding box of each vehicle, extract the vehicle image from the multimodal remote sensing image of the road area to obtain the multimodal remote sensing image of each vehicle in the road area.

[0099] Based on the location bounding box of each vehicle, vehicle images are extracted from the high-resolution optical remote sensing image and thermal infrared remote sensing image of the road area, respectively, to obtain the high-resolution optical remote sensing image and thermal infrared remote sensing image of each vehicle in the road area, that is, to obtain the multimodal remote sensing image of each vehicle in the road area.

[0100] Step S303: Based on the vehicle type and power type recognition model, identify the vehicle type and power type of each vehicle's multimodal remote sensing image.

[0101] Considering the differences in appearance between vehicles with different power types, for example, the charging port of new energy vehicles is usually on the front fender or rear bumper, while fuel vehicles do not have this feature. Also, new energy vehicles often have unique logos or streamlined designs on their bodies, while fuel vehicles may have large air intake grilles. Power type can be identified based on the differences in appearance between vehicles with different power types.

[0102] Considering the significant differences in heat distribution generated by the power systems of vehicles with different power types, for example, the combustion of internal combustion engines in gasoline vehicles generates a large amount of heat, which is concentrated in the engine compartment (front), while exhaust emissions heat the surrounding air, forming local high-temperature areas (near the exhaust pipe). The main heat sources of new energy vehicles are motors, batteries, and electronic devices (such as inverters), and the heat distribution is more dispersed (usually located in the chassis or rear), and there are no high-temperature areas heated by exhaust gases. Therefore, the power type can be identified based on the differences in heat distribution among vehicles with different power types.

[0103] High-resolution optical remote sensing images in multimodal vehicle remote sensing images present information such as the vehicle's appearance and structure, while thermal infrared remote sensing images reveal information such as heat distribution and heat sources. Combining these two different modalities of remote sensing images for vehicle type and power type identification can achieve complementary advantages and cross-validation between modalities, effectively improving the accuracy and robustness of vehicle type and power type identification.

[0104] The vehicle model and powertrain type recognition model in this embodiment is trained using multimodal remote sensing images of training vehicles (high-resolution optical remote sensing images and thermal infrared remote sensing images of the vehicles) labeled with vehicle model and powertrain type.

[0105] In one possible implementation, such as Figure 4 As shown, the vehicle type and power type recognition model in this embodiment may include: an optical image feature extraction module, a thermal infrared image feature extraction module, a cross-modal feature fusion module, and a multi-task classification module.

[0106] Specifically, for each vehicle's multimodal remote sensing image, the process of identifying the vehicle type and powertrain type based on the vehicle type and powertrain type recognition model can include:

[0107] Step a1-a: Based on the optical image feature extraction module, extract visual discrimination features from the high-resolution optical remote sensing image of the vehicle in the multimodal remote sensing image.

[0108] Among them, visual discrimination features represent information on the vehicle's exterior structure (such as body shape and outline lights) and component details (such as grille and wheel hubs).

[0109] Steps a1-b: Based on the thermal infrared image feature extraction module, extract thermal infrared identification features from the multimodal remote sensing images of the vehicle.

[0110] Among them, thermal infrared identification features characterize the temperature distribution and spatial layout of heat sources. Temperature distribution refers to the spatial variation of the radiation temperature on the vehicle surface (and near the surface), that is, the intensity and range of heat energy accumulation and diffusion in different parts of the vehicle. Spatial layout of heat sources refers to the location, shape, scale, and topological relationship of heat sources formed by major heat-generating components (such as the engine, battery pack, and exhaust pipe) in space.

[0111] Step a2: Based on the cross-modal feature fusion module, the extracted visual discrimination features are fused with the thermal infrared identification features to obtain cross-modal fusion features.

[0112] In one possible implementation, the process of fusing the extracted visual discriminative features with the thermal infrared identification features based on the cross-modal feature fusion module may include:

[0113] Step a21-a: Based on the cross-modal feature fusion module, enhanced thermal infrared identification features are generated by attention calculation with visual discrimination features as queries and thermal infrared identification features as keys and values.

[0114] Using visual discrimination features as queries and thermal infrared identification features as keys and values, the enhancement weights of the thermal infrared identification features are calculated. Based on the enhancement weights of the thermal infrared identification features, the thermal infrared identification features are weighted to obtain the enhanced thermal infrared identification features.

[0115] Steps a21-b: Based on the cross-modal feature fusion module, enhanced visual discriminative features are generated by attention calculation using the original thermal infrared identification features as queries and the visual discriminative features as keys and values.

[0116] Using the original thermal infrared identification features as the query and the visual discrimination features as the key and value, the enhancement weight of the visual discrimination features is calculated. Based on the enhancement weight of the visual discrimination features, the visual discrimination features are weighted to obtain the enhanced visual discrimination features.

[0117] Step a22: Based on the cross-modal feature fusion module, the enhanced thermal infrared identification features and the enhanced visual discrimination features are fused to obtain cross-modal fusion features.

[0118] Using visual discriminative features as the query vector and thermodynamic features as the key-value vector, cross-attention computation is used to enhance the thermal infrared identifier features. This process enables the model to adaptively focus on key thermal areas directly related to the powertrain (such as the engine high-temperature zone and exhaust pipe heat traces) based on prior knowledge of the vehicle's shape and component layout, while suppressing background thermal noise generated by the environment, sunlight, or non-critical components. Using thermal infrared identifier features as the query vector and visual discriminative features as the key-value vector, cross-attention computation is used to enhance the visual discriminative features. This process utilizes the spatial distribution and intensity information of heat sources to guide the model to focus on structural regions corresponding to heat-generating components (such as the air intake grille, charging port cover, and motor compartment outline), thereby strengthening the visual cues most relevant to powertrain type identification. The enhanced visual discriminative features and enhanced thermal infrared identifier features are then fused to obtain the final cross-modal fusion features. These cross-modal fusion features contain both discriminative appearance and structural information and essential thermodynamic information (such as thermal radiation distribution), laying a solid data foundation for subsequent high-precision and robust identification of vehicle models and powertrain types.

[0119] Step a3: Based on the multi-task classification module and cross-modal fusion features, predict the vehicle model and power type.

[0120] Specifically, the multi-task classification module includes a vehicle model classification module and a powertrain type classification module. The cross-modal fusion features are input into the vehicle model classification module and the powertrain type classification module respectively. The vehicle model classification module predicts the probability distribution of the vehicle in each vehicle model based on the fused features. Specifically, the vehicle model classification module predicts the probability distribution of the vehicle in each vehicle model based on the appearance structure information (such as macroscopic shape, structural proportions and component layout information) contained in the cross-modal fusion features. Then, the vehicle model is determined based on the probability distribution of the vehicle in each vehicle model. Similarly, the powertrain type classification module predicts the probability distribution of the vehicle in each powertrain type based on the cross-modal fusion features. Specifically, the powertrain type classification module predicts the probability distribution of the vehicle in each powertrain type based on the appearance structure information, component detail information, temperature distribution information and heat source spatial layout information contained in the cross-modal fusion features. Then, the powertrain type of the vehicle is determined based on the probability distribution of the vehicle in each powertrain type.

[0121] In some embodiments of this application, another implementation method is described for detecting vehicles and identifying the vehicle type and power type of each detected vehicle based on multimodal remote sensing images of a road area. In this implementation method, the multimodal remote sensing images of the road area may include registered high-resolution optical remote sensing images and thermal infrared remote sensing images, as well as hyperspectral optical remote sensing images, such as... Figure 5 As shown, the process for identifying vehicle model and powertrain type includes:

[0122] Step S501: Based on the high-resolution optical remote sensing image and thermal infrared remote sensing image in the multimodal remote sensing image of the road area, detect vehicles and identify the vehicle type and power type of each detected vehicle.

[0123] The specific implementation process of step S501 can be found in steps S301 to S303, which will not be repeated here in this embodiment.

[0124] It should be noted that multimodal remote sensing images may also include high-resolution optical remote sensing images or thermal infrared remote sensing images, as well as hyperspectral optical remote sensing images. Therefore, based on the high-resolution optical remote sensing images or thermal infrared remote sensing images in the multimodal remote sensing images of the road area, vehicles can be detected and the vehicle type and power type of each detected vehicle can be identified.

[0125] Step S502: By performing spectral analysis on the atmosphere above the road in the hyperspectral remote sensing image of the multimodal remote sensing image of the road area, the concentration distribution information of exhaust characteristic gases above the road area is generated.

[0126] Among them, exhaust characteristic gases refer to gases emitted into the atmosphere by fuel vehicles through exhaust gas that can be effectively detected by hyperspectral remote sensing images, such as CO2 and NOx. These gases can serve as "fingerprints" or "tracers" of fuel vehicle operation.

[0127] Exhaust gas characteristics are characterized by emission specificity and remote sensing detectability. Emission specificity refers to the fact that exhaust gases are mainly generated and emitted by fuel vehicles (internal combustion engines) during combustion, while new energy vehicles emit virtually no emissions or have negligible emissions during normal operation. Remote sensing detectability refers to the fact that exhaust gases have unique and stable spectral absorption or emission characteristics in the atmosphere, which can be effectively identified and quantitatively retrieved by spaceborne, airborne, or ground-based hyperspectral sensors in specific bands.

[0128] This embodiment uses atmospheric correction and spectral analysis of hyperspectral remote sensing images to invert the concentration of specific exhaust gas characteristic gases in the atmosphere above the road area, generating a concentration distribution map of exhaust gas characteristic gases. This concentration distribution map reflects the spatial pattern of near-ground pollution plumes formed by traffic activity emissions.

[0129] Step S503: Based on the power type identification results of each detected vehicle, and in conjunction with the concentration distribution information, determine the final power type of each detected vehicle.

[0130] Specifically, the process of determining the final power type of each detected vehicle based on the power type identification result of each vehicle and in conjunction with the concentration distribution information may include: determining the local gas concentration at the location of each detected vehicle by spatially registering the concentration distribution information with the location of each detected vehicle; and determining the final power type of each detected vehicle based on the power type identification result of each detected vehicle and the local gas concentration at the location of each detected vehicle.

[0131] More specifically, for each detected vehicle, if the vehicle's power type identification result indicates that the vehicle is a gasoline vehicle, and the local gas concentration at the vehicle's location is higher than a preset first concentration threshold, then the vehicle is determined to be a gasoline vehicle. If the vehicle's power type identification result indicates that the vehicle is a new energy vehicle, and the local gas concentration at the vehicle's location is lower than a preset second concentration threshold (the second concentration threshold is lower than the first concentration threshold), then the vehicle is determined to be a new energy vehicle. If the vehicle's power type identification result indicates that the vehicle is a gasoline vehicle, and the local gas concentration at the vehicle's location is lower than the preset first concentration threshold, then the confidence level of the vehicle's power type identification result is low, and the vehicle's power type can be re-identified. If the vehicle's power type identification result indicates that the vehicle is a new energy vehicle, and the local gas concentration at the vehicle's location is higher than the preset second concentration threshold, then the confidence level of the vehicle's power type identification result is low, and the vehicle's power type can be re-identified, or the vehicle can be determined to be a gasoline vehicle.

[0132] For each road area in the target city, after detecting vehicles based on remote sensing images of that road area and identifying the vehicle type and power type of each vehicle, the number of vehicles of each vehicle type-power type combination in that road area can be counted, and the emission factor matching each vehicle type-power type combination in that road area can be obtained. Then, based on the number of vehicles of each vehicle type-power type combination in that road area and the emission factor matching each vehicle type-power type combination in that road area, the carbon emissions of each vehicle type-power type combination in that road area can be determined. Finally, based on the carbon emissions of vehicles of various vehicle type-power type combinations in each road area of ​​the target city, the traffic carbon emissions of the target city can be determined.

[0133] The process of determining the carbon emissions of each vehicle type-power type combination within a road area, based on the number of vehicles of each vehicle type-power type combination within that road area and the emission factor matched to each vehicle type-power type combination within that road area, may include:

[0134] Step b1: Obtain the time-scale average mileage and energy consumption per unit mileage for each vehicle type-power type combination in the road area from the urban vehicle driving sampling database pre-built for the target city.

[0135] The average mileage over a time scale can be, but is not limited to, daily or monthly average mileage.

[0136] A city vehicle driving sampling database is pre-constructed for the target city. This database includes the time-scale average mileage (e.g., daily average mileage) and energy consumption per unit mileage for various vehicle type-power type combinations in the target city. Sampling surveys can be conducted on vehicle mileage and energy consumption data in the target city based on vehicle type and power type. Then, based on the sampling survey data, the time-scale average mileage (e.g., daily average mileage) and energy consumption per unit mileage for each vehicle type-power type combination can be estimated.

[0137] When determining the carbon emissions of vehicles with each vehicle type-power type combination in the road area, the time-scale average mileage (e.g., daily average mileage) and energy consumption per unit mileage of vehicles with each vehicle type-power type combination in the road area are obtained from the urban vehicle driving sampling database constructed for the target city.

[0138] Step b2: Based on the number of vehicles of each type-power combination in the road area and the emission factor matched with each type-power combination in the road area, and in combination with the time-scale average mileage and energy consumption per unit mileage of each type-power combination in the road area, calculate the carbon emissions of each type-power combination in the road area.

[0139] For each vehicle type-power type combination in this road area, if the power type in that combination is a gasoline vehicle, the carbon emissions of the vehicle in that combination can be calculated based on the following formula:

[0140] ECO2=N×L×EF×FC (1).

[0141] Where N is the number of vehicles with this model-power type combination in the road area, L is the average daily mileage of vehicles with this model-power type combination, EF is the emission factor (fuel emission factor) matched with this model-power type combination, and FC is the fuel consumption per kilometer of vehicles with this model-power type combination.

[0142] For each vehicle type-power type combination in this road area, if the power type in that combination is a new energy vehicle, the carbon emissions of the vehicle in that combination can be calculated based on the following formula:

[0143] EEV=M×P×β×EC (2).

[0144] Where M is the number of vehicles with this vehicle type-power type combination in the road area, P is the average daily mileage of vehicles with this vehicle type-power type combination, β is the emission factor (regional power grid emission factor) matched with this vehicle type-power type combination, and EC is the energy consumption per kilometer of vehicles with this vehicle type-power type combination.

[0145] For example, in a road area there are 100 gasoline-powered cars. The average daily driving distance of these cars is 300 kilometers. The fuel consumption per kilometer is 0.08L. The emission factor (gasoline emission factor) for these cars is 2.3kgCO2 / L. Therefore, the daily carbon emissions of these 100 gasoline-powered cars would be 300km / day × 0.08L / km × 2.3kgCO2 / L × 100 cars = 5520kg.

[0146] For example, in a road area there are 20 fuel-powered trucks. The average daily driving distance of the fuel-powered trucks is 200 kilometers. The fuel consumption of the fuel-powered trucks is 0.35L per kilometer. The emission factor (diesel emission factor) matched with the fuel-powered trucks is 2.68kgCO2 / L. Then the daily carbon emissions of the 20 fuel-powered trucks are 200km × 0.35L / km × 2.68kgCO2 / L × 20 trucks = 3752kg.

[0147] For example, in a road area there are 30 new energy cars. The average daily driving distance of the new energy cars is 30 kilometers. The electricity consumption per kilometer of the new energy cars is 0.15 kWh. The emission factor (regional power grid emission factor) matched with the new energy cars is 0.58 kg CO2 / kWh. Then the daily carbon emission of the 30 new energy cars is 30km × 0.15 kWh / km × 0.58 kg CO2 / kWh × 30 cars = 78.3 kg.

[0148] After determining the carbon emissions of each vehicle type-power type combination in each road area of ​​the target city based on remote sensing images, the total vehicle carbon emissions of the target city can be determined based on the carbon emissions of each vehicle type-power type combination in each road area of ​​the target city. After obtaining the total vehicle carbon emissions of the target city, if traffic carbon emission benchmark data that matches the target city in time and space is obtained, the traffic carbon emissions calculated for the target city can be compared with the traffic carbon emission benchmark data. Then, based on the comparison results, the emission factors in the emission factor comparison table are adjusted to make the emission factor comparison table more suitable for the target city. The adjusted emission factor comparison table is used for subsequent determination of traffic carbon emissions.

[0149] The urban traffic carbon emission measurement method provided in this application fundamentally revolutionizes the accounting paradigm for urban traffic carbon emissions. First, remote sensing images of various road areas in the target city are acquired. Then, multimodal fusion technology, combined with a machine learning model, is used to perform vehicle detection on the remote sensing images. For each detected vehicle, both vehicle type (e.g., sedan, SUV, truck, bus) and power type (fuel, new energy) are accurately identified. Based on the identification results, the number of vehicles of each vehicle type-power type combination in each road area is automatically counted. Combined with emission factors matching each vehicle type-power type, the carbon emissions of each vehicle type-power type combination in each road area are accurately calculated. Finally, by summarizing and integrating the carbon emissions of various vehicle type-power type combinations in each road area of ​​the target city, a scientific and reliable total urban-level traffic carbon emission is obtained.

[0150] The core value of the urban transportation carbon emission measurement method provided in this application lies in its revolutionary shift from the traditional "extensive estimation" relying on macro-statistics and hypothetical extrapolation to "precise calculation" based on actual measurements of each vehicle's attributes. By combining remote sensing technology with artificial intelligence algorithms, it completely solves the problems of traditional methods such as "unclear visibility" (incomplete spatial coverage), "inaccurate data" (confusion between vehicle type and powertrain), and "inaccurate calculation." The accurate data generated by the urban transportation carbon emission measurement method provided in this application will become the core engine driving the green and low-carbon transformation of urban transportation.

[0151] This application also provides an embodiment of an urban transportation carbon emission measurement device, such as... Figure 6 As shown, the urban traffic carbon emission measurement device may include: a remote sensing image acquisition unit 601, a vehicle detection and identification unit 602, a vehicle quantity statistics unit 603, an emission factor acquisition unit 604, a road area carbon emission determination unit 605, and an urban traffic carbon emission determination unit 606.

[0152] The remote sensing image acquisition unit 601 is used to acquire remote sensing images of each road area in the target city.

[0153] The vehicle detection and identification unit 602 is used to detect vehicles and identify the vehicle type and power type of each detected vehicle based on remote sensing images of the road area for each road area.

[0154] The vehicle quantity counting unit 603 is used to count the number of vehicles of each vehicle type-power type combination in the road area based on the recognition results.

[0155] The emission factor acquisition unit 604 is used to acquire emission factors that match each vehicle type-power type combination in the road area.

[0156] The road area carbon emission determination unit 605 is used to determine the carbon emissions of each type of vehicle-power type combination in the road area based on the number of vehicles in each type of vehicle-power type combination and the emission factor matched with each type of vehicle-power type combination.

[0157] The urban traffic carbon emission determination unit 606 is used to determine the traffic carbon emissions of the target city based on the carbon emissions of vehicles of different vehicle type-power type combinations in each road area of ​​the target city.

[0158] In one possible implementation, the road area carbon emission determination unit 605, when determining the carbon emissions of each vehicle type-power type combination in the road area based on the number of vehicles in each vehicle type-power type combination in the road area and the emission factor matched with each vehicle type-power type combination in the road area, is specifically used for:

[0159] From a pre-built urban vehicle driving sampling database for the target city, obtain the time-scale average mileage and energy consumption per unit mileage for each vehicle type-power type combination within the road area;

[0160] The carbon emissions of each vehicle type-power type combination in the road area are calculated based on the number of vehicles of each type-power type combination in the road area and the emission factor matched with each type-power type combination in the road area, combined with the time-scale average mileage and energy consumption per unit mile of each type-power type combination in the road area.

[0161] In one possible implementation, the emission factor acquisition unit 604, when acquiring the emission factor matched with each vehicle type-power type combination in the road area, is specifically used for:

[0162] From a pre-built emission factor lookup table, the emission factors matching each vehicle type-power type combination in the road area are obtained, and the emission factors matching each vehicle type-power type combination are obtained. The emission factor lookup table contains multiple vehicle type-power type combinations and the emission factors corresponding to each of the multiple vehicle type-power type combinations.

[0163] In one possible implementation, when the vehicle detection and recognition unit 602 detects vehicles and identifies the vehicle type and powertrain type of each detected vehicle based on remote sensing images of a road area, it is specifically used for:

[0164] Vehicle detection is performed on remote sensing images of the road area to obtain multiple vehicle location bounding boxes;

[0165] Based on the bounding box of each vehicle's location, vehicle images are extracted from the remote sensing images of the road area to obtain the remote sensing image of each vehicle in the road area;

[0166] Based on the vehicle type and power type recognition model, the remote sensing images of each vehicle in the road area are used to identify the vehicle type and power type. The vehicle type and power type recognition model is trained in advance using remote sensing images of training vehicles labeled with vehicle type and power type.

[0167] In one possible implementation, the remote sensing image of the road area is a multimodal remote sensing image, which includes a registered high-resolution optical remote sensing image and a thermal infrared remote sensing image.

[0168] When the vehicle detection and recognition unit 602 performs vehicle detection on the remote sensing image of the road area and obtains multiple vehicle location bounding boxes, it is specifically used for:

[0169] Vehicle detection is performed on the high-resolution optical remote sensing images in the multimodal remote sensing images of the road area to obtain multiple vehicle location bounding boxes.

[0170] When the vehicle detection and recognition unit 602 extracts vehicle images from the remote sensing image of the road area based on the location bounding box of each vehicle, and obtains the remote sensing image of each vehicle in the road area, it is specifically used for:

[0171] Based on the location bounding box of each vehicle, vehicle images are extracted from the multimodal remote sensing images of the road area to obtain the multimodal remote sensing image of each vehicle in the road area.

[0172] In one possible implementation, the remote sensing image of each vehicle is a multimodal remote sensing image, which includes a registered high-resolution optical remote sensing image and a thermal infrared remote sensing image.

[0173] When the vehicle detection and identification unit 602 identifies the vehicle type and power type of each vehicle in the road area based on the vehicle type and power type identification model, it is specifically used for:

[0174] For the multimodal remote sensing images of each vehicle in the road area, the optical image feature extraction module based on the vehicle type and power type recognition model extracts visual discrimination features that characterize the appearance structure and component details from the high-resolution optical remote sensing images of the vehicle's multimodal remote sensing images.

[0175] Based on the vehicle model and power type identification model, the thermal infrared image feature extraction module extracts thermal infrared identification features that characterize the temperature distribution and spatial layout of heat sources from the multimodal remote sensing images of the vehicle.

[0176] The cross-modal feature fusion module based on the vehicle model and power type recognition model fuses the visual discrimination features with the thermal infrared identification features to obtain cross-modal fusion features;

[0177] The multi-task classification module based on the vehicle model and powertrain type recognition model predicts the vehicle model and powertrain type based on the cross-modal fusion features.

[0178] In one possible implementation, when the vehicle detection and recognition unit 602 fuses the visual discrimination features with the thermal infrared identification features in the cross-modal feature fusion module based on the vehicle model and power type recognition model to obtain the cross-modal fused features, it is specifically used for:

[0179] The cross-modal feature fusion module based on vehicle model and power type recognition model generates enhanced thermal infrared identification features by using visual discriminative features as queries and thermal infrared identification features as keys and values ​​through attention calculation. It then generates enhanced visual discriminative features by using the original thermal infrared identification features as queries and visual discriminative features as keys and values ​​through attention calculation. Finally, it fuses the enhanced thermal infrared identification features with the enhanced visual discriminative features to obtain cross-modal fused features.

[0180] In one possible implementation, the remote sensing image of the road area is a multimodal remote sensing image, which includes high-resolution optical remote sensing images and / or thermal infrared remote sensing images, as well as hyperspectral optical remote sensing images.

[0181] When the vehicle detection and recognition unit 602 detects vehicles and identifies the vehicle type and powertrain of each detected vehicle based on remote sensing images of a road area, it is specifically used for:

[0182] Based on high-resolution optical remote sensing images and / or thermal infrared remote sensing images of road areas, vehicles are detected and the vehicle type and power type of each detected vehicle are identified.

[0183] By performing spectral analysis on the atmosphere above the road in the hyperspectral optical remote sensing image of the multimodal remote sensing image of the road area, the concentration distribution information of the characteristic exhaust gas above the road area is generated.

[0184] Based on the power type identification results of each vehicle, and in conjunction with the concentration distribution information, the final power type of each vehicle is determined.

[0185] In one possible implementation, when the vehicle detection and identification unit 602 determines the final power type of each vehicle based on the power type identification result of each vehicle and in conjunction with the concentration distribution information, it is specifically used for:

[0186] By spatially registering the concentration distribution information with the location of each vehicle, the local gas concentration at the location of each vehicle can be determined.

[0187] Based on the power type identification results of each vehicle and the local gas concentration at the location of each vehicle, the final power type of each vehicle is determined.

[0188] In one possible implementation, the urban transportation carbon emission measurement device may also include an emission factor update unit.

[0189] The emission factor update unit is used to compare the traffic carbon emissions calculated for the target city with the traffic carbon emission benchmark data if a traffic carbon emission benchmark data that matches the target city in time and space is obtained. Based on the comparison results, the emission factors in the emission factor comparison table are adjusted so that the adjusted emission factor comparison table can be used for the subsequent calculation of traffic carbon emissions for the target city.

[0190] The urban traffic carbon emission measurement device provided in this application embodiment can automatically detect vehicles in the road area and automatically and accurately identify vehicle type and power type based on remote sensing images of the road area, and thereby realize refined carbon emission calculation.

[0191] This application also provides an electronic device, which may include at least one processor and a memory connected to the processor.

[0192] The processor may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application; the memory may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage.

[0193] The memory is used to store computer programs, and the processor is used to execute the computer programs so that the electronic device can implement the urban traffic carbon emission calculation method provided in the above embodiments.

[0194] This application also provides a computer storage medium that carries one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device is able to implement the urban traffic carbon emission calculation method provided in the above embodiments.

[0195] This application also provides a computer program product, including computer-readable instructions, which, when executed on an electronic device, enable the electronic device to implement the urban traffic carbon emission calculation method provided in the above embodiments.

[0196] It should also be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. In addition, in the device embodiment drawings provided in this application, the connection relationship between modules indicates that they have a communication connection, which can be implemented as one or more communication buses or signal lines.

[0197] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware, or it can be implemented by special-purpose hardware including application-specific integrated circuits, special-purpose CPUs, special-purpose memory, special-purpose components, etc. Generally, any function performed by a computer program can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be diverse, such as analog circuits, digital circuits, or special-purpose circuits. However, for this application, software program implementation is more often the preferred implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a readable storage medium, such as a computer floppy disk, USB flash drive, mobile hard disk, ROM, RAM, magnetic disk, or optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, training equipment, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0198] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product.

[0199] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, training device, or data center to another website, computer, training device, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a training device or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives (SSDs)).

Claims

1. A method for calculating carbon emissions from urban transportation, characterized in that, include: Acquire remote sensing images of each road area in the target city; For each road area: Based on remote sensing images of the road area, vehicles are detected and the vehicle type and power type of each detected vehicle are identified; based on the identification results, the number of vehicles of each vehicle type-power type combination in the road area is counted, and the emission factor matching each vehicle type-power type combination is obtained; based on the number of vehicles of each vehicle type-power type combination in the road area and the emission factor matching each vehicle type-power type combination, the carbon emissions of each vehicle type-power type combination in the road area are determined; The traffic carbon emissions of the target city are determined based on the carbon emissions of vehicles of different vehicle type and powertrain combinations in each road area of ​​the target city.

2. The method for calculating urban transportation carbon emissions according to claim 1, characterized in that, The step of determining the carbon emissions of each vehicle type-power type combination within the road area based on the number of vehicles of each vehicle type-power type combination within the road area and the emission factor matched with each vehicle type-power type combination includes: From a pre-constructed urban vehicle driving sampling database for the target city, obtain the time-scale average mileage and energy consumption per unit mileage for each vehicle type-power type combination within the road area; The carbon emissions of each vehicle type-power type combination in the road area are calculated based on the number of vehicles in each vehicle type-power type combination within the road area and the emission factor matched with each vehicle type-power type combination, combined with the time-scale average mileage and energy consumption per unit mileage of each vehicle type-power type combination in the road area.

3. The method for calculating urban transportation carbon emissions according to claim 1, characterized in that, The process of obtaining the emission factor matching each of the vehicle model-powertrain combinations includes: From a pre-constructed emission factor lookup table, the emission factor corresponding to each of the vehicle model-power type combinations is obtained, and the emission factor matching each of the vehicle model-power type combinations is obtained. The emission factor lookup table contains multiple vehicle model-power type combinations and emission factors corresponding to each of the multiple vehicle model-power type combinations.

4. The method for calculating urban transportation carbon emissions according to claim 1, characterized in that, The remote sensing image of the road area is used to detect vehicles and identify the vehicle type and powertrain of each detected vehicle, including: Vehicle detection is performed on the remote sensing image of the road area to obtain multiple vehicle location bounding boxes; Based on the location bounding box of each vehicle, vehicle images are extracted from the remote sensing images of the road area to obtain the remote sensing image of each vehicle in the road area; Based on the vehicle type and power type recognition model, the remote sensing images of each vehicle in the road area are used to identify the vehicle type and power type. The vehicle type and power type recognition model is trained in advance using remote sensing images of training vehicles labeled with vehicle type and power type.

5. The method for calculating urban transportation carbon emissions according to claim 4, characterized in that, The remote sensing images of the road area are multimodal remote sensing images, which include registered high-resolution optical remote sensing images and thermal infrared remote sensing images. The remote sensing image of the road area is used to detect vehicles, resulting in multiple vehicle location bounding boxes, including: Vehicle detection is performed on high-resolution optical remote sensing images in the multimodal remote sensing images of the road area to obtain multiple vehicle location bounding boxes; The step of extracting vehicle images from the remote sensing image of the road area based on the location bounding box of each vehicle to obtain the remote sensing image of each vehicle in the road area includes: Based on the location bounding box of each vehicle, vehicle images are extracted from the multimodal remote sensing images of the road area to obtain a multimodal remote sensing image of each vehicle in the road area.

6. The method for calculating urban transportation carbon emissions according to claim 4, characterized in that, The remote sensing image of each vehicle is a multimodal remote sensing image, which includes a registered high-resolution optical remote sensing image and a thermal infrared remote sensing image. The vehicle type and powertrain type recognition model identifies the vehicle type and powertrain type of each vehicle in the remote sensing image of the road area, including: For each vehicle in the road area, the optical image feature extraction module based on the vehicle type and power type recognition model extracts visual discrimination features that characterize the appearance structure and component details from the high-resolution optical remote sensing images of the vehicle in the multimodal remote sensing images. Based on the vehicle model and power type identification model, the thermal infrared image feature extraction module extracts thermal infrared identification features that characterize the temperature distribution and spatial layout of heat sources from the multimodal remote sensing images of the vehicle. The cross-modal feature fusion module based on the vehicle model and power type recognition model fuses the visual discrimination features with the thermal infrared identification features to obtain cross-modal fusion features; The multi-task classification module based on the vehicle model and powertrain type recognition model predicts the vehicle model and powertrain type based on the cross-modal fusion features.

7. The method for calculating urban transportation carbon emissions according to claim 6, characterized in that, The process of fusing the visual discrimination features with the thermal infrared identification features to obtain cross-modal fusion features includes: An enhanced thermal infrared identification feature is generated by using the visual discrimination feature as the query and the thermal infrared identification feature as the key and value for attention calculation. An enhanced visual discriminative feature is generated by using the original thermal infrared identification feature as the query and the visual discriminative feature as the key and value for attention calculation. The enhanced thermal infrared identification features are fused with the enhanced visual discrimination features to obtain cross-modal fusion features.

8. The method for calculating urban transportation carbon emissions according to claim 1, characterized in that, The remote sensing images of the road area are multimodal remote sensing images, which include high-resolution optical remote sensing images and / or thermal infrared remote sensing images, as well as hyperspectral optical remote sensing images. The remote sensing image of the road area is used to detect vehicles and identify the vehicle type and powertrain of each detected vehicle, including: Based on high-resolution optical remote sensing images and / or thermal infrared remote sensing images in the multimodal remote sensing images of the road area, vehicles are detected and the model and power type of each detected vehicle are identified. By performing spectral analysis on the atmosphere above the road in the hyperspectral optical remote sensing image of the multimodal remote sensing image of the road area, the concentration distribution information of exhaust characteristic gases above the road area is generated. Based on the power type identification result of each vehicle and in conjunction with the concentration distribution information, the final power type of each vehicle is determined.

9. The method for calculating urban transportation carbon emissions according to claim 8, characterized in that, The step of determining the final power type of each vehicle based on the power type identification result of each vehicle and in conjunction with the concentration distribution information includes: By spatially registering the concentration distribution information with the location of each vehicle, the local gas concentration at the location of each vehicle is determined; The final power type of each vehicle is determined based on the power type identification result of each vehicle and the local gas concentration at the location of each vehicle.

10. The method for calculating urban transportation carbon emissions according to claim 3, characterized in that, Also includes: If traffic carbon emission benchmark data that matches the target city in time and space is obtained, the traffic carbon emissions calculated for the target city will be compared with the traffic carbon emission benchmark data. The emission factors in the emission factor comparison table are adjusted based on the comparison results so that the adjusted emission factor comparison table can be used for the subsequent calculation of the target city's transportation carbon emissions.

11. An electronic device, characterized in that, It includes at least one processor and a memory connected to the processor, wherein: The memory is used to store computer programs; The processor is used to execute the computer program so that the electronic device can implement the urban traffic carbon emission calculation method as described in any one of claims 1 to 10.

12. A computer storage medium, characterized in that, The storage medium carries one or more computer programs, which, when executed by an electronic device, enable the electronic device to implement the urban traffic carbon emission calculation method as described in any one of claims 1 to 10.

Images