Method for predicting yield of pinus koraiensis cones

Abstract

The invention relates to a prediction method, in particular to a pinus koraiensis pine cone yield prediction method which comprises the following steps: S1, unmanned aerial vehicle oblique photography, S2, image processing, S3, orthographic splicing, S4, pine cone recognition and S5, yield prediction. According to the method, a large area can be covered in a short time, Korean pine image data can be rapidly acquired, indoor data analysis is carried out, the acquisition cost is only one tenth of that of traditional manual investigation, and almost no personnel safety risk exists. According to the method, the unmanned aerial vehicle is adopted to collect data, the efficiency is high, investigation can be carried out in a slightly allowed aerial photography window period in weather, the flexibility is high, the influence of topographic elements and accumulated snow elements is small, and the adaptability is high. The method comprehensively investigates the measurement area, identifies the positions and number of pinecones through an algorithm, greatly reduces manual intervention, and improves the investigation precision.

Description

Technical Field

[0001] The invention relates to a prediction method, in particular to a method for predicting the yield of Korean pine cones. Background Art

[0002] After the pine cones of Korean pine are peeled, they become pine seeds. Pine seeds are not only delicious, but also have nutritional and health functions, and there is a great market demand. The right to pick Korean pine cones usually creates huge economic value through auctions. At present, the assessment before the auction of picking rights adopts the traditional form of manual investigation by climbing up trees tens of meters high. However, the use of manual investigation will have the following problems: First, the traditional form of manual investigation by climbing up trees tens of meters high requires deep forests to collect one by one, which has extremely high time and labor costs and great safety risks; second, the traditional method is greatly affected by weather images, and it is difficult to conduct investigations on windy and snowy days; third, manual investigation and prediction can only be carried out in the form of sampling investigations, and the accuracy of the investigation mainly depends on the experience and judgment of the investigators, and the labor intensity is high; fourth, the traditional manual investigation is greatly affected by the terrain, and the investigation cost is affected by the accessibility of the survey area, the complexity of the terrain, and the degree of fragmentation, especially in the three northeastern provinces where Korean pine is traditionally planted, which is easily affected by snow accumulation; fourth, the traditional method can only rely on manual labeling and identification, which has low accuracy, low efficiency, high labor intensity, and investigators are prone to fatigue. Summary of the invention

[0003] The present invention provides a method for predicting the yield of Korean pine cones, the main purpose of which is to overcome the above-mentioned technical problems existing in the existing method for manually predicting the yield of Korean pine cones.

[0004] In order to solve the above technical problems, the present invention adopts the following technical solutions to achieve the above problems:

[0005] A method for predicting the yield of Korean pine cones, characterized in that it comprises the following steps:

[0006] S1. Drone oblique photography: In winter, when the cones of red pine are brown, the route is planned for the survey area. The pitch angle of the gimbal is set within the range of 20-45°. The route is set according to the forward / rearward or left / right view, and the drone oblique photography of the survey area is carried out.

[0007] S2. Image processing: Use image processing software to process the images taken by drones, highlight the red in the images, and filter out the green in the images;

[0008] S3, ortho stitching: align images from different perspectives to ensure they are precisely aligned to achieve higher accuracy;

[0009] S4, pine cone recognition: mark the red area in the registered image with RGB color R>200;

[0010] S5. Yield prediction: By counting the number of all markers, the yield of red pine in the survey area can be predicted.

[0011] Furthermore, in step S2, the specific method of image processing is: processing the image through general image processing software, so that the color characteristics of the pine cone can be significantly highlighted through processing, so that the recognition rate exceeds 90%, specifically adjusting the color balance item of the image, increasing cyan and blue, and reducing red and yellow.

[0012] Furthermore, in step S3, the orthophoto stitching process is as follows:

[0013] S31. Check the data after aerial photography to ensure that the geographical location information of the aerial photos is accurate;

[0014] S32, importing the checked data into general 3D modeling software;

[0015] S33, point cloud registration, aligning the point cloud data in images taken at different angles to form a complete point cloud model;

[0016] S34, generate DSM using point cloud data through TIN triangulation model;

[0017] S35, stitching the multiple images together according to the external orientation parameters and point cloud data of the images to generate a preliminary image stitching result;

[0018] S36. Based on DSM data, orthorectify the stitched images to eliminate the tilt and distortion in the images and generate true orthorectified images.

[0019] Further, in step S4, the process of pine cone identification is as follows:

[0020] S41. Define input image:

[0021] Input image data: Image(x,y), where (x,y) represents the pixel position of the image;

[0022] The color value of each pixel is Image(x,y) = (R(x,y), G(x,y), B(x,y)), which represents the value of the red, green, and blue channels respectively;

[0023] S42, setting red threshold: red threshold R = 200;

[0024] S43, mark the red area:

[0025] Traverse the image data and find the pixel positions where the red channel value is greater than the threshold:

[0026]

[0027] Generate a set of locations marked in red:

[0028]

[0029] Through the above steps, the area where the red channel value is greater than 200 in the image can be marked.

[0030] Furthermore, in step S5, the process of production prediction is as follows:

[0031] S51, determine the area of ​​each pixel:

[0032] Assume that the spatial resolution of the image is Resolution, that is, the actual area represented by each pixel. Usually, this value is one of the metadata of the image. Let Apixe l is the area of ​​each pixel;

[0033] S52. Count the number of pixels in the red area:

[0034] Count the total number of pixels marked red:

[0035]

[0036] S53. Calculate the total area of ​​the red area:

[0037] The total number of pixels with a total area of ​​red multiplied by the area of ​​each pixel:

[0038] A total =N red ×A pixel ;

[0039] Through the above steps, the total area of ​​the red area in the image can be calculated, thereby predicting the yield of red pine cones.

[0040] From the above description of the present invention, it can be seen that compared with the prior art, the present invention has the following advantages: 1. This method can cover a large area in a short period of time, quickly collect Korean pine image data, and conduct indoor data analysis. The acquisition cost is only a few tenths of the traditional manual survey, and there is almost no risk to personnel safety. 2. This method uses drones to collect data with high efficiency, and the survey can be carried out as long as the weather allows for aerial photography. It has strong flexibility, is less affected by terrain factors and snow factors, and has strong adaptability. 3. This method conducts a comprehensive survey of the survey area, and identifies the location and number of pine cones through algorithms, which greatly reduces manual intervention and improves survey accuracy. DETAILED DESCRIPTION

[0041] A method for predicting the yield of Korean pine cones, characterized in that it comprises the following steps:

[0042] S1. Drone oblique photography: In winter, when the cones of red pine are brown, the route is planned for the survey area. The pitch angle of the gimbal is set within the range of 20-45°. The route is set according to the forward / rearward or left / right view, and the drone oblique photography of the survey area is carried out.

[0043] This photography method can collect image data at different angles, and the workload of data collection and post-processing is only 40% of that of traditional oblique photography. While meeting the data acquisition accuracy, the work efficiency is greatly improved. Moreover, aerial photography is carried out in the winter period (December of the current year to April of the following year) when the characteristics of pine cones and red pine needles are the most different, which effectively improves the accuracy of yield prediction.

[0044] S2. Image processing: Use image processing software to process the images taken by drone photography, highlight the red in the images, and filter out the green in the images;

[0045] S3, ortho stitching: align images from different perspectives to ensure they are accurately aligned to achieve higher accuracy;

[0046] S4, pine cone recognition: mark the red area in the registered image with RGB color R>200;

[0047] S5. Yield prediction: By counting the number of all markers, the yield of red pine in the survey area can be predicted.

[0048] Specifically, in step S2, the specific method of image processing is: processing the image through general image processing software (such as Photoshop), so that the color characteristics of the pine cone can be significantly highlighted through processing, so that the recognition rate exceeds 90%, specifically adjusting the color balance item of the image, increasing cyan and blue, and reducing red and yellow.

[0049] Specifically, in step S3, the orthophoto stitching process is as follows:

[0050] S31. Check the data after aerial photography to ensure that the geographical location information of the aerial photos is accurate;

[0051] S32. Import the checked data into general 3D modeling software (such as Pix4D, ContextCapture, Agisoft Metashape, DJI Zhitu, etc.);

[0052] S33, point cloud registration, aligning the point cloud data in images taken at different angles to form a complete point cloud model;

[0053] S34, generating a DSM (digital surface model) using the point cloud data through the TIN triangulation model;

[0054] S35, stitching the multiple images together according to the external orientation parameters and point cloud data of the images to generate a preliminary image stitching result;

[0055] S36. Based on DSM data, orthorectify the stitched images to eliminate the tilt and distortion in the images and generate true orthorectified images.

[0056] Specifically, in step S4, the process of pine cone identification is as follows:

[0057] S41. Define input image:

[0058] Input image data: Image(x,y), where (x,y) represents the pixel position of the image;

[0059] The color value of each pixel is Image(x,y) = (R(x,y), G(x,y), B(x,y)), which represents the value of the red, green, and blue channels respectively;

[0060] S42, setting the red threshold: red threshold R = 200;

[0061] S43, mark the red area:

[0062] Traverse the image data and find the pixel positions where the red channel value is greater than the threshold:

[0063]

[0064] Generate a set of locations marked in red:

[0065]

[0066] Through the above steps, the area where the red channel value is greater than 200 in the image can be marked.

[0067] This recognition method can automatically identify and mark the characteristic parts of pine cones, realize automated operations, and has high work efficiency and precision, laying the foundation for the later pine cone harvesting.

[0068] Specifically, in step S5, the process of production prediction is as follows:

[0069] S51, determine the area of ​​each pixel:

[0070] Assume that the spatial resolution of the image is Resolution, that is, the actual area represented by each pixel. Usually, this value is one of the metadata of the image. Let Apixe l is the area of ​​each pixel;

[0071] S52. Count the number of pixels in the red area:

[0072] Count the total number of pixels marked red:

[0073]

[0074] S53. Calculate the total area of ​​the red area:

[0075] The total number of pixels with a total area of ​​red multiplied by the area of ​​each pixel:

[0076] A total =N red ×A pixel ;

[0077] Through the above steps, the total area of ​​the red area in the image can be calculated, thereby predicting the yield of red pine cones.

[0078] The above is only a specific implementation of the present invention, but the design concept of the present invention is not limited thereto. Any non-substantial changes to the present invention using this concept shall be deemed as an infringement of the protection scope of the present invention.

Claims

1. A method for predicting the yield of Korean pine cones, characterized in that: The following steps are involved: S1. Drone oblique photography: In winter, when the cones of red pine are brown, the route is planned for the survey area. The pitch angle of the gimbal is set within the range of 20-45°. The route is set according to the forward / rearward or left / right view, and the drone oblique photography of the survey area is carried out. S2. Image processing: Use image processing software to process the images taken by drones, highlight the red in the images, and filter out the green in the images; S3, ortho stitching: align images from different perspectives to ensure they are precisely aligned to achieve higher accuracy; S4, pine cone recognition: mark the red area in the registered image with RGB color R>200; S5. Yield prediction: By counting the number of all markers, the yield of red pine in the survey area can be predicted.

2. A method for predicting the yield of Korean pine cones as claimed in claim 1, characterized in that: In step S2, the specific method of image processing is: processing the image through general image processing software, so that the color characteristics of the pine cone can be significantly highlighted through processing, so that the recognition rate exceeds 90%, specifically adjusting the color balance item of the image, increasing cyan and blue, and reducing red and yellow.

3. A method for predicting the yield of Korean pine cones as claimed in claim 1, characterized in that: In step S3, the process of ortho stitching is as follows: S31. Check the data after aerial photography to ensure that the geographical location information of the aerial photos is accurate; S32, importing the checked data into general 3D modeling software; S33, point cloud registration, aligning the point cloud data in images taken at different angles to form a complete point cloud model; S34, generate DSM using point cloud data through TIN triangulation model; S35, stitching the multiple images together according to the external orientation parameters and point cloud data of the images to generate a preliminary image stitching result; S36. Based on DSM data, orthorectify the stitched images to eliminate the tilt and distortion in the images and generate true orthorectified images.

4. A method for predicting the yield of Korean pine cones as claimed in claim 1, characterized in that: In step S4, the process of pine cone identification is as follows: S41. Define input image: Input image data: Image(x,y), where (x,y) represents the pixel position of the image; The color value of each pixel is Image(x,y) = (R(x,y), G(x,y), B(x,y)), which represents the value of the red, green, and blue channels respectively; S42, setting red threshold: red threshold R = 200; S43, mark the red area: Traverse the image data and find the pixel positions where the red channel value is greater than the threshold: Generate a set of locations marked in red: Through the above steps, the area where the red channel value is greater than 200 in the image can be marked.

5. A method for predicting the yield of Korean pine cones as claimed in claim 1, characterized in that: In step S5, the process of production prediction is as follows: S51, determine the area of ​​each pixel: Assume that the spatial resolution of the image is Resolution, that is, the actual area represented by each pixel. Usually, this value is one of the metadata of the image. Let A pixel is the area of ​​each pixel; S52. Count the number of pixels in the red area: Count the total number of pixels marked red: S53. Calculate the total area of ​​the red area: The total number of pixels with a total area of ​​red multiplied by the area of ​​each pixel: A total =N red ×A pixel ; Through the above steps, the total area of ​​the red area in the image can be calculated, thereby predicting the yield of red pine cones.

Images