Agglomerate fog forecasting method based on improved double-brightness difference method and grey forecasting model

Abstract

The invention relates to an agglomerate fog forecasting method based on an improved double-brightness difference method and a grey forecasting model, which belongs to the field of agglomerate fog forecasting and comprises the following steps of: (1) establishing a video image data set; (2) graying the image; (3) image filtering and noise reduction; (4) feature region segmentation and extraction; (5) calculating visibility; (6) repeating the steps (2) to (5) to obtain a time sequence of visibility; and (7) inputting a visibility time sequence, and forecasting the agglomerate fog based on a grey prediction model. The method has the advantages that the improved double-brightness difference method is combined with the grey prediction model, the problem that the monitoring range is limited due to the current situations that meteorological equipment is sparsely distributed, equipment cost is high, maintenance is complex and the like in an existing agglomerate fog forecasting method is solved, and timely and accurate forecasting of agglomerate fog in road traffic is achieved.

Description

Technical Field

[0001] The invention relates to a fog cluster prediction method based on an improved double brightness difference method and a grey prediction model, and belongs to the field of fog cluster prediction. Background Art

[0002] Fog is a meteorological phenomenon that is highly sudden and has a small impact range, but it poses a great threat to traffic safety. In meteorology, visibility is usually used as an important parameter for fog forecasting. Traditional fog forecasting methods mainly rely on visibility observation instruments and lidar and other meteorological instruments, but they have limitations such as sparse equipment distribution, high equipment cost, and complex maintenance, which results in a limited monitoring range and makes it difficult to capture the local and instantaneous characteristics of fog, and cannot meet the real-time and accurate requirements of fog forecasting. In recent years, the application of video surveillance and image processing technology has made it possible to capture the real-time dynamics of fog occurrence, but different algorithms will affect the accuracy of visibility detection results, and it is impossible to monitor and forecast the development trend of fog. Summary of the invention

[0003] In order to solve the above problems existing in the prior art, the present invention proposes a fog cluster prediction method based on an improved double brightness difference method and a grey prediction model. The collected road monitoring video data are analyzed by using improved image processing technology to obtain a visibility time series, which is input into the GM (1,1) grey prediction model of the oscillation sequence to predict the visibility change trend over time, realize the prediction of fog clusters, and improve the safety of road traffic.

[0004] The present invention provides a method for predicting fog clusters based on an improved double brightness difference method and a grey prediction model. The technical solution of the present invention is:

[0005] A method for predicting fog clusters based on an improved double brightness difference method and a grey prediction model comprises the following steps: (1) establishing a video image data set; (2) graying the image; (3) filtering and denoising the image; (4) segmenting and extracting feature regions; (5) calculating visibility; (6) repeating steps (2) to (5) to obtain a time series of visibility; and (7) inputting the visibility time series to predict fog clusters based on a grey prediction model.

[0006] The step (1) is specifically as follows: decomposing the video data collected by the road monitoring equipment frame by frame to form a video image data set.

[0007] The step (2) is specifically as follows:

[0008] 2.1 Read the color image in the video image data set obtained in step (1);

[0009] 2.2 The red R, green G and blue B channels of each pixel in the color image are processed into a grayscale single channel through the weighted average formula: GRAY = 0.30r + 0.59g + 0.11b

[0010] 2.3. Generate a grayscale image and realize image grayscale.

[0011] The step (3) is specifically:

[0012] Use a median filter to filter and reduce noise on the grayscale image obtained in step (2).

[0013] In the step (4), the feature region is segmented using the region growing method, which specifically includes the following steps:

[0014] 4.1. Select a pixel in the sky area of ​​the image obtained in step (3) as a seed;

[0015] 4.2 Merge similar pixels in the neighborhood into the seed area through the threshold to expand the neighborhood;

[0016] 4.3 Repeat steps 4.1 and 4.2, merging the new pixel as a seed with its neighborhood until all pixels are included in the feature region to achieve region segmentation;

[0017] 4.4 Take the average grayscale of each pixel in the segmented sky area to obtain the grayscale Gt of the target object.

[0018] The step (5) specifically includes the following steps:

[0019] Select two targets, whose grayscales can be obtained from step (4), and are denoted as Gt1 and Gt2 respectively; the distances from the target to the camera lens are L 1 and L 2 , L 2 -L 1 Obtained through the road marking distance, substitute into the formula: Among them, V d is visibility, G g1 and G g2 are the background sky brightness of the two targets respectively; in order to reduce the test error, the target in the selected area is consistent with the line of sight of the observation point, so that G g1 =G g2 , and thus calculate the visibility V d .

[0020] In the step (6), the following steps are specifically included:

[0021] Repeat steps (2) to (5) to calculate the visibility of all images in the video image dataset, and extract the time information of each image to obtain the time series of visibility.

[0022] In the step (7), the following steps are specifically included: the time series of visibility obtained in step (6) is input into the GM(1,1) grey prediction model based on the oscillation sequence, and the generating columns, matrix B and data vector Y are constructed respectively, and the time response sequence of the oscillation sequence GM(1,1) grey prediction model is obtained as follows: This enables the visibility to be predicted and analyzed, and the forecast of fog to be achieved.

[0023] The advantages of the present invention are: using an improved double brightness difference method combined with a gray prediction model, the problems existing in the existing fog cluster forecasting methods, such as sparse distribution of meteorological equipment, high equipment cost and complex maintenance, which lead to limited monitoring range, difficulty in capturing the local and instantaneous characteristics of fog clusters, and unsatisfactory accuracy of visibility detection results, making it impossible to monitor and forecast the development trend of fog clusters, thereby achieving timely and accurate forecasting of fog clusters in road traffic. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 The present invention is a flow chart of the method for forecasting fog clusters.

[0025] Figure 2 This is a schematic diagram of selecting seeds in a characteristic region using the regional growing method in the method for predicting fog clusters of the present invention.

[0026] Figure 3 This is a diagram showing the regional segmentation effect of the fog forecasting method of the present invention.

[0027] Figure 4 The present invention provides a method for predicting fog using a road surface marking distance measurement diagram.

[0028] Figure 5 This is a forecast result diagram of the grey prediction model of the fog cluster forecasting method of the present invention. DETAILED DESCRIPTION

[0029] The present invention will be further described below in conjunction with specific embodiments, and the advantages and features of the present invention will become clearer as the description proceeds. However, these embodiments are exemplary only and do not constitute any limitation to the scope of the present invention. It should be understood by those skilled in the art that the details and forms of the technical solution of the present invention may be modified or replaced without departing from the spirit and scope of the present invention, but these modifications and replacements all fall within the scope of protection of the present invention.

[0030] See also Figures 1 to 5The present invention relates to a method for predicting fog clusters based on an improved double brightness difference method and a gray prediction model, comprising the following steps: (1) establishing a video image data set; (2) graying the image; (3) filtering and denoising the image; (4) segmenting and extracting feature regions; (5) calculating visibility; (6) repeating steps (2) to (5) to obtain a time series of visibility; and (7) inputting the visibility time series to predict fog clusters based on the gray prediction model.

[0031] Based on the settings in the above steps, the following are achieved:

[0032] Improved accuracy: By combining the double brightness difference method and the grey prediction model, the formation and development of fog can be captured and predicted more accurately.

[0033] Real-time data processing: Through the establishment and real-time processing of video image data sets, it is possible to respond to environmental changes in a timely manner and provide timely fog forecasts.

[0034] High degree of automation: Automated image processing and data analysis processes reduce manual intervention and improve the efficiency and objectivity of forecasts.

[0035] Full data utilization: Through steps such as grayscale conversion, filtering and noise reduction, and feature area segmentation, the information in the video image data is fully mined and utilized.

[0036] Strong environmental adaptability: This method is suitable for different environmental and climatic conditions and has strong adaptability and universal applicability.

[0037] Good prediction timeliness: Through time series analysis, it is possible to provide forecasts on the development trend of fog clusters, which helps to take preventive measures in advance.

[0038] The step (1) specifically comprises: decomposing the video data collected by the road monitoring equipment frame by frame to form a video image data set.

[0039] The step (2) is specifically as follows:

[0040] 2.1 Read the color image in the video image data set obtained in step (1);

[0041] 2.2 The three channels of red R, green G and blue B of each pixel in the color image are processed into a single grayscale GRAY channel through the weighted average formula: GRAY = 0.30R + 0.59G + 0.11B;

[0042] 2.3 Generate a grayscale image and realize image grayscale.

[0043] The step (3) specifically includes: using a median filter to perform filtering and noise reduction processing on the grayscale image obtained in step (2).

[0044] In the step (4), the feature region is segmented using the region growing method, which specifically includes the following steps:

[0045] 4.1 As Figure 2 As shown, a pixel point is selected as a seed (red box) in the sky area of ​​the image obtained in step (3);

[0046] 4.2 Merge similar pixels in the neighborhood into the seed region through the threshold to expand the neighborhood; 4.3 Repeat steps 4.1 and 4.2, merge the new pixel as a seed with its neighborhood, until all pixels are included in the feature region, and achieve regional segmentation, such as Figure 3 As shown; 4.4 takes the average grayscale of each pixel in the segmented sky area to obtain the grayscale Gt of the target object.

[0047] The step (5) specifically includes the following steps:

[0048] like Figure 4 As shown, two targets are selected, whose grayscales can be obtained by step (4), and are denoted as Gt1 and Gt2 respectively; the distances from the target to the camera lens are L1 and L2 respectively, and L2-L1 is obtained by the road marking distance, and is substituted into the formula: Among them, Vd is visibility, Gg1 and Gg 2 are the background sky brightness of the two targets respectively; in order to reduce the test error, the target in the selected area is consistent with the line of sight of the observation point, so that G g1 =G g2 , and thus the visibility Vd is calculated.

[0049] In the step (6), the following steps are specifically included:

[0050] Repeat steps (2) to (5) to calculate the visibility of all images in the video image dataset, and extract the time information of each image to obtain the time series of visibility.

[0051] In the step (7), the following steps are specifically included: the time series of visibility obtained in step (6) is input into the GM(1,1) grey prediction model based on the oscillation sequence, and the generating columns, matrix B and data vector Y are constructed respectively, and the time response sequence of the oscillation sequence GM(1,1) grey prediction model is obtained as follows: Thus, the visibility can be predicted and analyzed, and the fog can be predicted. The prediction results are as follows: Figure 5 shown.

[0052] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for predicting fog clusters based on an improved double brightness difference method and a grey prediction model, characterized in that: The following steps are involved: (1) Establish a video image dataset; (2) Grayscale the image; (3) Filter and reduce the noise of the image; (4) Segment and extract the feature region; (5) Calculate the visibility; (6) Repeat steps (2) to (5) to obtain the visibility time series; (7) Input the visibility time series and realize the prediction of fog clusters based on the grey prediction model.

2. The method for predicting fog clusters based on an improved double brightness difference method and a grey prediction model according to claim 1, characterized in that: The step (1) is specifically as follows: decomposing the video data collected by the road monitoring equipment frame by frame to form a video image data set.

3. The method for predicting fog clusters based on an improved double brightness difference method and a grey prediction model according to claim 1, characterized in that: The step (2) is specifically as follows: 2.1 Read the color image in the video image data set obtained in step (1); 2.2 The red R, green G and blue B channels of each pixel in the color image are processed into a grayscale single channel through the weighted average formula: GRAY = 0.30R + 0.59G + 0.11B 2.

3. Generate a grayscale image and realize image grayscale.

4. The method for predicting fog clusters based on an improved double brightness difference method and a grey prediction model according to claim 1, characterized in that: The step (3) is specifically: Use a median filter to filter and reduce noise on the grayscale image obtained in step (2).

5. The method for predicting fog clusters based on an improved double brightness difference method and a grey prediction model according to claim 1, characterized in that: In the step (4), the feature region is segmented using the region growing method, which specifically includes the following steps: 4.

1. Select a pixel in the sky area of ​​the image obtained in step (3) as a seed; 4.2 Merge similar pixels in the neighborhood into the seed area through the threshold to expand the neighborhood; 4.3 Repeat steps 4.1 and 4.2, merging the new pixel as a seed with its neighborhood until all pixels are included in the feature region to achieve region segmentation; 4.4 Take the average grayscale of each pixel in the segmented sky area to obtain the grayscale Gt of the target object.

6. The method for predicting fog clusters based on an improved double brightness difference method and a grey prediction model according to claim 1, characterized in that: The step (5) specifically includes the following steps: Select two targets, whose grayscales can be obtained from step (4), and are denoted as Gt1 and Gt2 respectively; the distances from the targets to the camera lens are L1 and L2 respectively, and L2-L1 is obtained by the road marking distance and substituted into the formula: Among them, V d is visibility, G g1 and G g2 are the background sky brightness of the two targets respectively; in order to reduce the test error, the target in the selected area is consistent with the line of sight of the observation point, so that G g1 =G g2 , and thus calculate the visibility V d .

7. The method for predicting fog clusters based on an improved double brightness difference method and a grey prediction model according to claim 1, characterized in that: The step (6) specifically includes the following steps: Repeat steps (2) to (5) to calculate the visibility of all images in the video image dataset, and extract the time information of each image to obtain the time series of visibility.

8. The method for predicting fog clusters based on an improved double brightness difference method and a grey prediction model according to claim 1, characterized in that: In the step (7), the following steps are specifically included: the time series of visibility obtained in step (6) is input into the GM(1,1) grey prediction model based on the oscillation sequence, and the generating columns, matrix B and data vector Y are constructed respectively, and the time response sequence of the oscillation sequence GM(1,1) grey prediction model is obtained as follows: This enables the visibility to be predicted and analyzed, and the forecast of fog to be achieved.

Images