A carbon emission analysis system for rail transit construction

Abstract

The present invention relates to the field of data analysis, and in particular to a carbon emission analysis system for rail transit construction. The present invention sets a dynamic acquisition module, an environmental screening module and a feature analysis module. The dynamic acquisition module is used to collect image data of a construction site. The environmental screening module is used to divide the feature explicit categories of image segments. The feature analysis module is used to analyze the image segments, including performing a global analysis on each video frame of the image segment to determine whether a predetermined target object exists, or extracting video frames from feature image sub-segments and non-feature image sub-segments in different extraction methods, and only analyzing the feature areas of each of the video frames to determine whether a predetermined target object exists. The present invention can identify carbon emission sources in large-area construction sites, reduce image data redundancy when processing a large amount of image data, and ensure target recognition accuracy while reducing the consumption of image processing computing power.

Description

Technical Field

[0001] The present invention relates to the field of data analysis, and in particular to a carbon emission analysis system for rail transit construction. Background Art

[0002] As the global climate change problem becomes increasingly serious, reducing carbon emissions has become a common concern of governments and society. All walks of life have begun to pay attention to carbon emissions during manufacturing, production and construction. For example, carbon emissions are easily generated during construction, and it is necessary to continuously detect carbon emission sources and optimize construction strategies. Therefore, relevant data collection and carbon emission source identification are crucial.

[0003] For example, China Patent Publication No.: CN114627387A, a regional carbon emission prediction method, device and medium based on Beidou positioning and remote sensing image detection, including: obtaining satellite remote sensing images of the target area, identifying the types of buildings and the number of each type of buildings; using the carbon emission prediction model of each building to predict the current carbon emissions of the building; collecting the position information of each motor vehicle in the target area based on Beidou positioning to obtain its speed, and estimating the carbon emissions of each motor vehicle in the target area according to the speed of the motor vehicle and the energy emission factor, energy net calorific value, and characteristic factor; using the hierarchical analysis method to determine the weight of all buildings and motor vehicles in the solution layer to the regional carbon emissions; calculating the carbon emissions in the target area according to the carbon emissions and weights of the buildings and motor vehicles in the target area. This invention can effectively solve the problem of obtaining carbon emission data in a large area, and is flexible in application and relatively low in cost.

[0004] However, there are still the following problems in the prior art:

[0005] Using image analysis to inspect large-scale construction sites will generate a lot of image data. Analyzing the image data consumes a lot of computing power. In addition, due to the influence of the shooting environment, there is a lot of redundancy in the collected images, which can easily interfere with the accuracy of target feature recognition. Summary of the invention

[0006] To this end, the present invention provides a carbon emission analysis system for rail transit construction, which is used to overcome the problem that the image analysis method is used in the prior art to detect large-area construction sites, which will generate a lot of image data, consume a lot of computing power to analyze the image data, and due to the influence of the shooting environment, there is a lot of redundancy in the collected images, which can easily interfere with the accuracy of target feature recognition.

[0007] To achieve the above objectives, the present invention provides a carbon emission analysis system for rail transit construction, which includes:

[0008] A dynamic acquisition module, comprising a drone flying along a predetermined route and an image acquisition unit disposed on the drone for acquiring image data;

[0009] An environment screening module, which is connected to the dynamic acquisition module, and is used to calculate feature fusion parameters based on the chromaticity of the local area and the object contour distribution characteristics in each video frame of the image segment, so as to classify the feature dominant categories of the image segment, wherein the feature dominant categories include dominant categories and non-dominant categories;

[0010] A feature analysis module, which is connected to the environment screening module, is used to analyze the image fragments according to the feature explicit categories of the image fragments, including:

[0011] Performing global analysis on each video frame of the image segment to determine whether a predetermined target object exists;

[0012] Or, identifying characteristic image sub-segments according to differences in basic features in adjacent video frames of the image segment, extracting video frames from the characteristic image sub-segments and non-characteristic image sub-segments in different extraction methods, identifying characteristic regions according to chromaticity gradients between local regions and adjacent regions in each of the video frames, and analyzing only the characteristic regions of each of the video frames to determine whether a predetermined target object exists;

[0013] The basic characteristics include chromaticity and brightness, and the predetermined target object is a pre-set object that can generate carbon emissions.

[0014] Furthermore, the environment screening module calculates the feature fusion parameter according to formula (1):

[0015]

[0016] In formula (1), D represents a feature fusion parameter, n represents the number of local regions in a video frame, Pi represents the average chromaticity of the i-th local region in the video frame, P0 represents the average chromaticity of the video frame, Ne represents the number of object contours, Ne0 represents a preset number threshold, Ng represents the average spacing between the center points of object contours, Ng0 represents a preset spacing threshold, α represents a chromaticity weight threshold, and β represents a contour weight threshold.

[0017] Furthermore, the environment screening module compares the feature fusion parameter corresponding to the image segment with a preset feature fusion parameter threshold to classify the feature dominant category of the image segment.

[0018] If the feature fusion parameter is greater than or equal to the feature fusion parameter threshold, the environment screening module determines that the image segment is a dominant category;

[0019] If the feature fusion parameter is less than the feature fusion parameter threshold, the environment screening module determines that the image segment is a hidden category.

[0020] Furthermore, the feature analysis module is also used to select a method for analyzing the image fragment according to the feature explicit category of the image fragment, including:

[0021] If the image segment is of the dominant category, the feature analysis module selects to perform a global analysis on each video frame of the image segment to determine whether there is a predetermined target object;

[0022] If the image segment is of a hidden category, the feature analysis module selects to analyze only the feature area of ​​each video frame to determine whether a predetermined target object exists.

[0023] Furthermore, the feature analysis module divides the image segment into a plurality of sub-segments, and calculates the difference parameter of each sub-segment according to formula (2):

[0024]

[0025] In formula (2), E represents a difference parameter, ns represents the number of video frames in a sub-segment, Ps(i) represents the average chroma of the i-th video frame in the sub-segment, Ps(i-1) represents the average chroma of the i-1-th video frame in the sub-segment, Ps0 represents the average chroma of each video frame in the sub-segment, Ls(i) represents the average brightness of the i-th video frame in the sub-segment, Ls(i-1) represents the average brightness of the i-1-th video frame in the sub-segment, and Ls0 represents the average brightness of each video frame in the sub-segment.

[0026] Furthermore, the feature analysis module compares the difference parameter of each sub-segment with a preset difference parameter threshold to identify the feature image sub-segment, wherein:

[0027] If the difference parameter of the sub-segment is less than the difference parameter threshold, the feature parsing module identifies the sub-segment as a feature image sub-segment.

[0028] Furthermore, the feature analysis module is also used to select a method for extracting video frames, wherein:

[0029] Randomly extracting a single video frame from a feature image sub-segment;

[0030] Extract all video frames in non-featured image sub-segments.

[0031] Furthermore, the feature analysis module calculates the chromaticity gradient of each local area and each adjacent area of ​​the video frame respectively.

[0032] If the chromaticity gradient between the local area of ​​the video frame and any adjacent area is greater than a predetermined gradient threshold, the feature analysis module identifies the local area as a feature area;

[0033] The chromaticity gradient is the difference between the chromaticity value of the local area and the chromaticity value of the adjacent area.

[0034] Furthermore, an image processing model is provided in the feature analysis module to identify a predetermined target object in a video frame.

[0035] Furthermore, the feature analysis module is also used to store video frames containing predetermined target objects.

[0036] Compared with the prior art, the present invention sets a dynamic acquisition module, an environmental screening module and a feature analysis module. The dynamic acquisition module is used to collect image data of the construction site, the environmental screening module is used to divide the feature explicit categories of the image fragments, and the feature analysis module is used to analyze the image fragments, including performing a global analysis on each video frame of the image fragment to determine whether there is a predetermined target object, or extracting video frames from feature image sub-fragments and non-feature image sub-fragments in different extraction methods, and only analyzing the feature areas of each of the video frames to determine whether there is a predetermined target object. The present invention can identify carbon emission sources in large-area construction sites, reduce image data redundancy when processing a large amount of image data, and ensure target recognition accuracy while reducing the consumption of image processing computing power.

[0037] In particular, since the image data acquisition environment is outdoor, the image data collected over a large area contains some image fragments that do not have effective features. Due to the large amount of data, it will consume more computing power if each image frame is analyzed. Therefore, the present invention calculates feature fusion parameters and considers the chromaticity of the local area and the distribution of object contours. In actual situations, the image data collected outdoors may contain a large number of background features, such as ground images. The chromaticity difference of the local area of ​​this type of image data is small, and the distribution of object contours is small. Therefore, the present invention calculates feature fusion parameters and divides the feature-dominant categories of the image fragments. The chromaticity and contour features of the image are easy to obtain and can be quickly calculated, thereby providing data support for the subsequent targeted analysis of the video frames, thereby reducing image data redundancy when processing a large amount of image data, and ensuring target recognition accuracy while reducing the consumption of image processing computing power.

[0038] In particular, the present invention performs global analysis on each video frame of the image segment of the explicit category. In actual situations, the image segment of the explicit category may contain more effective feature information, so performing global analysis can reduce feature omissions and ensure recognition accuracy.

[0039] In particular, the present invention identifies characteristic image sub-segments for image segments of non-explicit categories. The similarity of each video frame in the characteristic image sub-segment is high, and a single video frame has data representation of the entire characteristic image sub-segment. Therefore, for the characteristic image sub-segment, only a single video frame is randomly extracted for analysis. In addition, the present invention also identifies characteristic areas based on the chromaticity gradient between the local area and the adjacent area. In actual situations, the chromaticity of the effective feature will be different from the surrounding environment. Therefore, the present invention identifies the characteristic area and only analyzes the characteristic area, thereby reducing image data redundancy when processing a large amount of image data, and ensuring target recognition accuracy while reducing the consumption of image processing computing power. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is a structural block diagram of a carbon emission analysis system for rail transit construction according to an embodiment of the invention;

[0041] Figure 2 It is a logic block diagram of image segment category determination according to an embodiment of the present invention;

[0042] Figure 3 A logic block diagram of identifying feature image sub-segments according to an embodiment of the invention;

[0043] Figure 4 The figure is a logic block diagram of identifying feature areas according to an embodiment of the present invention. DETAILED DESCRIPTION

[0044] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0045] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0046] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the term "connection" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be an indirect connection through an intermediate medium, or it can be the internal communication of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0047] See also Figures 1 to 4 As shown, Figure 1 This is a structural block diagram of a carbon emission analysis system for rail transit construction according to an embodiment of the invention. Figure 2This is a logic block diagram of image segment category determination according to an embodiment of the invention. Figure 3 The following is a logic block diagram of identifying feature image sub-segments according to an embodiment of the present invention. Figure 4 The figure is a logic block diagram of identifying feature areas according to an embodiment of the present invention.

[0048] A carbon emission analysis system for rail transit construction according to an embodiment of the present invention includes:

[0049] A dynamic acquisition module, comprising a drone flying along a predetermined route and an image acquisition unit disposed on the drone for acquiring image data;

[0050] An environment screening module, which is connected to the dynamic acquisition module, and is used to calculate feature fusion parameters based on the chromaticity of the local area and the object contour distribution characteristics in each video frame of the image segment, so as to classify the feature dominant categories of the image segment, wherein the feature dominant categories include dominant categories and non-dominant categories;

[0051] A feature analysis module, which is connected to the environment screening module, is used to analyze the image fragments according to the feature explicit categories of the image fragments, including:

[0052] Performing global analysis on each video frame of the image segment to determine whether a predetermined target object exists;

[0053] Or, identifying characteristic image sub-segments according to differences in basic features in adjacent video frames of the image segment, extracting video frames from the characteristic image sub-segments and non-characteristic image sub-segments in different extraction methods, identifying characteristic regions according to chromaticity gradients between local regions and adjacent regions in each of the video frames, and analyzing only the characteristic regions of each of the video frames to determine whether a predetermined target object exists;

[0054] The basic characteristics include chromaticity and brightness, and the predetermined target object is a pre-set object that can generate carbon emissions.

[0055] Specifically, the present invention does not limit the specific structure of the image acquisition unit. The image acquisition unit can be a photographic device configured on the drone, which will not be described in detail.

[0056] Specifically, the present invention does not limit the specific structures of the environment screening module and the feature analysis module, both of which can be composed of logic components or a combination of logic components, and the logic components include field programmable components, computers and microprocessors.

[0057] Specifically, the present invention does not limit the method of extracting chromaticity, brightness and object contours in an image. Chromaticity and brightness are basic parameters of image data and are easy to obtain. For the recognition of object contours, an algorithm or model that can recognize the contours of objects in images can be pre-trained and imported into the logic component to implement the corresponding function. This is a prior art and will not be elaborated on.

[0058] Specifically, an image processing model is provided in the feature analysis module to identify predetermined target objects in video frames. The present invention does not limit the specific form of the image processing model. To identify specific objects in images, a target detection model in the field of computer vision can be used, such as the YOLO model, which realizes real-time target detection by converting the target detection problem into a regression problem. A convolutional neural network model can also be used, such as the VGGNet model, which has a deeper network structure and extracts higher-level features through multiple convolutional layers and fully connected layers. These are all prior arts and will not be described in detail here.

[0059] Specifically, the predetermined target objects are pre-set objects that can generate carbon emissions. After identification, they provide data support for construction managers, which can help construction managers manage construction sites, optimize construction methods, and thus reduce carbon emissions;

[0060] For the predetermined target object, technicians in this field can set it according to the needs, for example, a generator set, a mixing station, etc. Of course, it is necessary to cooperate with a pre-set image processing module and use the image of the predetermined target object as a training set to train the corresponding image processing model to achieve the function of identifying the predetermined target object. This will not be repeated here.

[0061] Specifically, the environment screening module calculates the feature fusion parameter according to formula (1):

[0062]

[0063] In formula (1), D represents a feature fusion parameter, n represents the number of local regions in a video frame, Pi represents the average chromaticity of the i-th local region in the video frame, P0 represents the average chromaticity of the video frame, Ne represents the number of object contours, Ne0 represents a preset number threshold, Ng represents the average spacing between the center points of object contours, Ng0 represents a preset spacing threshold, α represents a chromaticity weight threshold, β represents a contour weight threshold, and in this embodiment, α is 0.6 and β is 0.4.

[0064] Specifically, the environment screening module compares the feature fusion parameter corresponding to the image segment with a preset feature fusion parameter threshold to classify the feature dominant category of the image segment.

[0065] If the feature fusion parameter is greater than or equal to the feature fusion parameter threshold, the environment screening module determines that the image segment is a dominant category;

[0066] If the feature fusion parameter is less than the feature fusion parameter threshold, the environment screening module determines that the image segment is a hidden category.

[0067] Specifically, in this embodiment, the feature fusion parameter threshold D0 is pre-set, wherein a number of image segments can be collected as samples at the construction site through the dynamic acquisition module in advance, and the average value ΔD of the feature fusion parameters corresponding to each image segment is solved, and D0 is set to ΔD×g, g represents the accuracy coefficient, 1.15<g<1.3.

[0068] Since the image data acquisition environment is outdoor, the image data collected over a large area contains some image fragments that do not have effective features. Due to the large amount of data, it will consume more computing power if each image frame is analyzed. Therefore, the present invention calculates feature fusion parameters and considers the chromaticity of the local area and the distribution of object contours. In actual situations, the image data collected outdoors may contain a large number of background features, such as ground images. The chromaticity difference of the local area of ​​this type of image data is small, and the distribution of object contours is small. Therefore, the present invention calculates feature fusion parameters and divides the feature-dominant categories of the image fragments. The chromaticity and contour features of the image are easy to obtain and can be quickly calculated, thereby providing data support for the subsequent targeted analysis of the video frames, thereby reducing image data redundancy when processing a large amount of image data, and ensuring target recognition accuracy while reducing the consumption of image processing computing power.

[0069] Specifically, the feature analysis module is further used to select a method for analyzing the image segment according to the feature explicit category of the image segment, including:

[0070] If the image segment is of the dominant category, the feature analysis module selects to perform a global analysis on each video frame of the image segment to determine whether there is a predetermined target object;

[0071] If the image segment is of a hidden category, the feature analysis module selects to analyze only the feature area of ​​each video frame to determine whether a predetermined target object exists.

[0072] The present invention performs global analysis on each video frame of the image segment of the dominant category. In actual situations, the image segment of the dominant category may contain more effective feature information, so the global analysis is performed to reduce feature omissions and ensure recognition accuracy.

[0073] Specifically, the feature analysis module divides the image segment into several sub-segments, and calculates the difference parameters of each sub-segment according to formula (2):

[0074]

[0075] In formula (2), E represents a difference parameter, ns represents the number of video frames in a sub-segment, Ps(i) represents the average chroma of the i-th video frame in the sub-segment, Ps(i-1) represents the average chroma of the i-1-th video frame in the sub-segment, Ps0 represents the average chroma of each video frame in the sub-segment, Ls(i) represents the average brightness of the i-th video frame in the sub-segment, Ls(i-1) represents the average brightness of the i-1-th video frame in the sub-segment, and Ls0 represents the average brightness of each video frame in the sub-segment.

[0076] Specifically, the feature analysis module compares the difference parameter of each sub-segment with a preset difference parameter threshold to identify the feature image sub-segment, wherein:

[0077] If the difference parameter of the sub-segment is less than the difference parameter threshold, the feature parsing module identifies the sub-segment as a feature image sub-segment.

[0078] Specifically, in this embodiment, in order to ensure the difference of video frames in the sub-segments, the difference parameter threshold is set within the interval [0.3, 0.8].

[0079] Specifically, the feature analysis module is also used to select a method for extracting video frames, wherein:

[0080] Randomly extracting a single video frame from a feature image sub-segment;

[0081] Extract all video frames in non-featured image sub-segments.

[0082] Specifically, the feature analysis module calculates the chromaticity gradient of each local area and each adjacent area of ​​the video frame respectively.

[0083] If the chromaticity gradient between the local area of ​​the video frame and any adjacent area is greater than a predetermined gradient threshold, the feature analysis module identifies the local area as a feature area;

[0084] The chromaticity gradient is the difference between the chromaticity value of the local area and the chromaticity value of the adjacent area.

[0085] Specifically, in this embodiment, the gradient threshold H0 is pre-determined, wherein a number of image segments containing target features are obtained, and the chromaticity gradient of the local area where the target feature is located and each adjacent area in the video frame is calculated to solve the average chromaticity gradient of the local area and the adjacent area in a single video frame, and the average of the chromaticity gradient averages is solved and recorded as ΔH, and H0 is set to 1.2×ΔH.

[0086] The present invention identifies characteristic image sub-segments for image segments of non-explicit categories. The similarity of each video frame in the characteristic image sub-segment is high, and a single video frame has data representation of the entire characteristic image sub-segment. Therefore, for the characteristic image sub-segment, only a single video frame is randomly extracted for analysis. In addition, the present invention also identifies characteristic areas based on the chromaticity gradient between the local area and the adjacent area. In actual situations, the chromaticity of the effective feature will be different from the surrounding environment. Therefore, the present invention identifies the characteristic area and only analyzes the characteristic area, thereby reducing image data redundancy when processing a large amount of image data, and ensuring target recognition accuracy under the premise of reducing the consumption of image processing computing power.

[0087] Specifically, the feature analysis module is also used to store video frames containing predetermined target objects.

[0088] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

Claims

1. A carbon emission analysis system for rail transit construction, characterized in that: include: A dynamic acquisition module, comprising a drone flying along a predetermined route and an image acquisition unit disposed on the drone for acquiring image data; An environment screening module, which is connected to the dynamic acquisition module, and is used to calculate feature fusion parameters based on the chromaticity of the local area and the object contour distribution characteristics in each video frame of the image segment, so as to classify the feature dominant categories of the image segment, wherein the feature dominant categories include dominant categories and non-dominant categories; A feature analysis module, which is connected to the environment screening module, is used to analyze the image fragments according to the feature explicit categories of the image fragments, including: Performing global analysis on each video frame of the image segment to determine whether a predetermined target object exists; Or, identifying characteristic image sub-segments according to differences in basic features in adjacent video frames of the image segment, extracting video frames from the characteristic image sub-segments and non-characteristic image sub-segments in different extraction methods, identifying characteristic regions according to chromaticity gradients between local regions and adjacent regions in each of the video frames, and analyzing only the characteristic regions of each of the video frames to determine whether a predetermined target object exists; The basic characteristics include chromaticity and brightness, and the predetermined target object is a pre-set object that can generate carbon emissions; The environment screening module calculates the feature fusion parameters according to formula (1): In formula (1), D represents a feature fusion parameter, n represents the number of local regions in a video frame, Pi represents the average chromaticity of the i-th local region in the video frame, P0 represents the average chromaticity of the video frame, Ne represents the number of object contours, Ne0 represents a preset number threshold, Ng represents the average spacing between the center points of object contours, Ng0 represents a preset spacing threshold, α represents a chromaticity weight threshold, and β represents a contour weight threshold; The environment screening module compares the feature fusion parameter corresponding to the image segment with a preset feature fusion parameter threshold to classify the feature dominant category of the image segment. If the feature fusion parameter is greater than or equal to the feature fusion parameter threshold, the environment screening module determines that the image segment is a dominant category; If the feature fusion parameter is less than the feature fusion parameter threshold, the environment screening module determines that the image segment is a hidden category; The feature analysis module is also used to select a method for analyzing the image segment according to the feature explicit category of the image segment, including: If the image segment is of the dominant category, the feature analysis module selects to perform a global analysis on each video frame of the image segment to determine whether there is a predetermined target object; If the image segment is a hidden category, the feature analysis module selects to analyze only the feature area of ​​each video frame to determine whether there is a predetermined target object; The feature analysis module divides the image segment into several sub-segments and calculates the difference parameters of each sub-segment according to formula (2): In formula (2), E represents the difference parameter, ns represents the number of video frames in the sub-segment, Ps(i) represents the average chroma of the i-th video frame in the sub-segment, Ps(i-1) represents the average chroma of the i-1-th video frame in the sub-segment, Ps0 represents the average chroma of each video frame in the sub-segment, Ls(i) represents the average brightness of the i-th video frame in the sub-segment, Ls(i-1) represents the average brightness of the i-1-th video frame in the sub-segment, and Ls0 represents the average brightness of each video frame in the sub-segment.

2. The carbon emission analysis system for rail transit construction according to claim 1 is characterized in that: The feature analysis module compares the difference parameter of each sub-segment with a preset difference parameter threshold to identify the feature image sub-segment, wherein: If the difference parameter of the sub-segment is less than the difference parameter threshold, the feature parsing module identifies the sub-segment as a feature image sub-segment.

3. The carbon emission analysis system for rail transit construction according to claim 1 is characterized in that: The feature analysis module is also used to select a method for extracting video frames, wherein: Randomly extracting a single video frame from a feature image sub-segment; Extract all video frames in non-featured image sub-segments.

4. The carbon emission analysis system for rail transit construction according to claim 1 is characterized in that: The feature analysis module calculates the chromaticity gradient of each local area and each adjacent area of ​​the video frame respectively. If the chromaticity gradient between the local area of ​​the video frame and any adjacent area is greater than a predetermined gradient threshold, the feature analysis module identifies the local area as a feature area; The chromaticity gradient is the difference between the chromaticity value of the local area and the chromaticity value of the adjacent area.

5. The carbon emission analysis system for rail transit construction according to claim 1 is characterized in that: The feature analysis module is provided with an image processing model for identifying a predetermined target object in a video frame.

6. The carbon emission analysis system for rail transit construction according to claim 1 is characterized in that: The feature analysis module is also used to store the video frames containing the predetermined target objects.

Images