A corn cross-section phenotype parameter detection method based on image processing and polar coordinate expansion

Abstract

The application discloses a kind of corn cross section phenotypic parameter detection method based on image processing and polar coordinate development, belong to agricultural phenotype detection technical field.It includes: obtaining the image containing corn cross section and reference information reflecting scale;Extract the outer contour and inner core contour of cross section and carry out pairing, extract the inside and outside diameter of pixel;Calculate the conversion ratio of pixel to physical size;With cross section as reference, the rectangular image band is obtained by spatial transformation to the region of interest thereof;Extract one-dimensional distribution signal along radial projection;Obtain the optimal period of signal by autocorrelation analysis, and obtain the initial peak set by using matched filtering peak search;Based on annular feature, the gap of possible missed detection is interpolated and supplemented;Finally, the number of rows of corn ear is determined by applying biological even row constraint, which is used to assist corn yield determination, and the real phenotypic parameter is calculated by combining conversion ratio.The application is light in calculation, strong in noise resistance, fully combines agricultural breeding logic, and can accurately output physical level phenotype data.

Description

Technical Field

[0001] This invention relates to the field of agricultural phenotypic parameter detection technology, and more specifically, to a method for detecting maize cross-sectional phenotypic parameters based on image processing and polar coordinate expansion. Background Technology

[0002] Maize is one of the world's most important food crops, and accurate yield assessment is of paramount strategic importance for ensuring food security. Phenotypic parameters on the cross-section of the maize ear (such as the number of rows, actual kernel width, and actual length) are key indicators for assessing maize yield potential and selecting superior varieties. Traditionally, the measurement of these parameters has relied primarily on manual observation and caliper measurements, which is not only time-consuming and labor-intensive but also easily influenced by subjective judgment, making it difficult to meet the needs of modern large-scale automated screening.

[0003] With the development of computer vision technology, several image processing-based methods for corn phenotypic detection have been proposed. Traditional image processing methods have significant advantages such as fast operation speed, lightweight deployment, and no need for large amounts of data annotation, making them very suitable for portable agricultural equipment with limited computing resources. However, relying solely on traditional operators often struggles to stably extract boundaries when faced with complex corn cross-sectional shapes, uneven lighting, or deformation. Although heavy-duty deep learning techniques have been introduced in recent years for direct dense detection or pixel-level segmentation, this typically requires extremely high annotation costs and massive GPU computing power, making rapid and lightweight deployment difficult. Moreover, regardless of whether it's a traditional method or deep learning, corn cross-sections may exhibit "missing kernels" due to poor development or minor physical damage. If corn kernels are treated as ordinary, unrelated, independent targets, it is difficult to effectively infer and correct for missing kernel regions, leading to row counting errors.

[0004] Furthermore, due to differences in camera shooting distance and focal length, the dimensions extracted directly from images are usually only in pixel units, failing to directly reflect the true biological physical scale. Moreover, many existing image detection methods often overlook the crucial agricultural yield measurement logic that "the number of grains in a cross-section equals the number of rows in a corn ear." In summary, there is a need for a lightweight detection method that can adapt to changes in corn morphology and fully utilize its biological characteristics, accurately extracting physical dimensions while incorporating biological constraints to assist in yield measurement. Summary of the Invention

[0005] To address the above-mentioned deficiencies or improvement needs of existing technologies, this invention provides an automatic detection method for maize cross-sectional phenotypic parameters based on image processing and polar coordinate expansion. It aims to solve the problems of missing physical conversion of parameters, high computing power requirements for high-throughput processing, and errors in missing kernel counting in existing technologies, and to achieve high-throughput, high-precision extraction of maize phenotypic parameters and auxiliary yield measurement.

[0006] To achieve the above objectives, this invention provides an automatic detection method for maize cross-sectional phenotypic parameters based on image processing and polar coordinate expansion, comprising the following steps: S1. Acquire cross-sectional images of corn, and extract the outer contour and inner core contour of each corn cross-section in the image; at the same time, obtain scale reference information reflecting the physical scale of the shooting; S2. Pair the outer contour and inner core contour of the same corn entity, and extract the outer diameter and inner diameter that reflect the pixel size of the corn cross section; at the same time, calculate the conversion ratio from pixel to physical size according to the scale reference information. S3. Using the same corn entity as a reference, the region of interest between the outer diameter and the inner diameter is spatially transformed and expanded to obtain a rectangular expanded image band. S4. Project the rectangular unfolded image strip radially to obtain a one-dimensional signal reflecting the distribution characteristics of corn kernels or gaps between corn kernels. S5. Perform periodic analysis on the one-dimensional signal to obtain the optimal period, and obtain an initial set of peak values ​​based on the optimal period through peak detection. Each peak value corresponds to the position of a corn kernel or the gap between corn kernels. S6. Based on the optimal period and the annular closed characteristic of the corn cross section, interpolation is performed to fill any gaps that may be missed in the initial peak set. S7. Apply a biological even number constraint to the total number of peak values ​​after the leak filling treatment to determine the final number of corn kernels in the cross section. The number of kernels in the cross section represents the number of rows in the corn ear and is used to assist in the evaluation of corn yield. Combined with the conversion ratio, the outer diameter and inner diameter are converted into actual physical dimensions, and then the phenotypic parameters such as the physical length and physical width of the corn kernels are calculated.

[0007] The present invention also provides an electronic device, comprising: a computer-readable storage medium and a processor; The computer-readable storage medium is used to store executable instructions; The processor is used to read executable instructions stored in the computer-readable storage medium and execute the above-described automatic detection method for corn cross-sectional phenotypic parameters based on image processing and polar coordinate expansion.

[0008] The present invention also provides a computer-readable storage medium storing computer instructions for causing a processor to execute the above-described automatic detection method for corn cross-sectional phenotypic parameters based on image processing and polar coordinate expansion.

[0009] In summary, compared with the prior art, the above-described technical solutions conceived by this invention can achieve the following beneficial effects: 1. Obtaining absolute physical dimensions: The automatic detection method for corn cross-sectional phenotypic parameters based on image processing and polar coordinate expansion provided by this invention introduces scale reference information, which bridges the gap between pixels and real physical space, and can directly and accurately output specific millimeter-level data.

[0010] 2. Computational Lightweight and High-Throughput Processing: The automatic detection method for maize cross-sectional phenotypic parameters based on image processing and polar coordinate expansion provided by this invention cleverly combines contour extraction with traditional one-dimensional signal processing technology, giving full play to the advantages of traditional methods such as fast running speed and low overhead. It is very suitable for large-scale real-time deployment in portable field devices or high-throughput phenotyping pipelines.

[0011] 3. Feature dimensionality reduction and intelligent correction: Polar coordinate expansion is introduced to transform the complex ring distribution into a one-dimensional signal; and through biological even-number row constraints and periodic interpolation, effective compensation is performed when there are missing maize.

[0012] 4. Deep integration with yield measurement applications: The automatic detection method for maize cross-sectional phenotypic parameters based on image processing and polar coordinate expansion provided by this invention clarifies that the counting results of maize kernels in the cross-section essentially reflect the "row count" of the whole ear of maize, directly connecting with downstream agricultural yield measurement models. Attached Figure Description

[0013] Figure 1 This is an overall flowchart of a detection method provided in an embodiment of the present invention.

[0014] Figure 2 This is a visualization of the entire process of corn cross-section unfolding from polar coordinates to one-dimensional signal peak finding and omission compensation in an embodiment of the present invention.

[0015] Figure 3 This is a demonstration of the detection results of an embodiment of the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0017] This invention provides an automatic detection method for maize cross-sectional phenotypic parameters based on image processing and polar coordinate expansion. The overall flowchart is as follows: Figure 1 As shown, it includes the following steps: S1. Acquire cross-sectional images of corn, and extract the outer contour and inner core contour of each corn cross-section in the image; at the same time, obtain scale reference information reflecting the physical scale of the shooting; S2. Pair the outer contour and inner core contour of the same corn entity, and extract the outer diameter and inner diameter that reflect the pixel size of the corn cross section; at the same time, calculate the conversion ratio from pixel to physical size according to the scale reference information. S3. Using the same corn entity as a reference, the region of interest between the outer diameter and the inner diameter is spatially transformed and expanded to obtain a rectangular expanded image band. S4. Project the rectangular unfolded image strip radially to obtain a one-dimensional signal reflecting the distribution characteristics of corn kernels or gaps between corn kernels. S5. Perform periodic analysis on the one-dimensional signal to obtain the optimal period, and obtain an initial set of peak values ​​based on the optimal period through peak detection. Each peak value corresponds to the position of a corn kernel or the gap between corn kernels. S6. Based on the optimal period and the annular closed characteristic of the corn cross section, interpolation is performed to fill any gaps that may be missed in the initial peak set. S7. Apply a biological even number constraint to the total number of peak values ​​after the leak filling treatment to determine the final number of corn kernels in the cross section. The number of kernels in the cross section represents the number of rows in the corn ear and is used to assist in the evaluation of corn yield. Combined with the conversion ratio, the outer diameter and inner diameter are converted into actual physical dimensions, and then the phenotypic parameters such as the physical length and physical width of the corn kernels are calculated.

[0018] Example 1 This embodiment will demonstrate the entire process of corn cross-section unfolding from polar coordinates to one-dimensional signal peak finding and omission compensation, and its visualization effect is shown in the figure below. Figure 2 As shown, the test results are as follows Figure 3 As shown.

[0019] Step S1, Image Acquisition and Scale Reference Information Acquisition: Acquire raw images of the cross-section of a corn cob using an industrial camera or portable device, and extract the outer contour bounding box of the corn and the inner contour bounding box of the corn cob. This invention provides two alternative implementation methods to reflect scale reference information: Example A (Same-plane Reference Object Method): During photography, a known physical diameter is placed on the same plane as the corn cross-section. Using a reference object (e.g., a 25.0 mm coin), the system extracts the pixel boundaries of the coin using the YOLO model. Example B (Depth Sensor Inference Algorithm): When it's inconvenient to place the coin in the actual environment, the system utilizes the built-in depth sensor of a smartphone or industrial camera (such as a LiDAR or ToF lens), or is based on the camera's focal length. and focus distance It directly obtains the absolute physical distance from the camera to the corn cross-section, and then calculates the physical scale represented by the pixel through an optical geometric imaging model.

[0020] S2. Inner and outer frame pairing and physical conversion ratio calculation: Calculate the center point coordinates of each bounding box, and use nearest neighbor search to pair inner and outer bounding boxes whose distance is less than a preset threshold. For Example A, obtain the pixel diameter of the coin bounding box. Calculate the conversion ratio from pixel to physical size. (mm / pixel). For Example B, directly using the optical focal length formula. get Take half the pixel size of both the outer frame and the inner frame as the pixel outer diameter. and pixel inner diameter .

[0021] Step S3: Adaptive Spatial Transformation Unfolding. To adapt to corn cross-sections of different shapes, this step adopts an adaptive unfolding strategy: evaluating the shape characteristics of the paired outer contours and calculating the ratio of their major axis to minor axis. Alternative Scheme 1 (Standard Circular Unfolding): If the difference between the major and minor axes is less than a preset deformation threshold, it is considered a standard circle. Using the center point... Perform a transformation from standard polar coordinates to rectangular coordinates with the origin as the reference point. Let the width of the unfolded image band be... (Typically, 360 or more finer sample numbers are used), height is For any pixel in the target image The formula for the corresponding original image coordinates (x, y) is: (1) The original image pixels are obtained through bilinear interpolation, completing the ring-to-rectangular unfolding mapping. Alternative scheme 2 (elliptical unfolding): If the difference between the major and minor axes is significant due to compression (i.e., considered an ellipse), then an ellipse-polar coordinate transformation is performed: based on the major axis... and short axis Formula for dynamically adjusting the sampling radius (2) To avoid nonlinear stretching distortion of the image in the polar regions of the ellipse.

[0022] Step S4: One-dimensional signal projection. Let the expanded region of interest image matrix be... ,in Representative perspective, This represents the effective radial core region. The system calculates the average pixel grayscale value along the radial direction: (3) To make the gap stand out as a peak, the signal is reversed: (4) Obtain a one-dimensional distributed signal .

[0023] Step S5: Autocorrelation period estimation and peak finding for one-dimensional signals Perform mean-reduction processing (i.e.) ), and calculate the autocorrelation function: : (5) The optimal permutation period is determined by detecting the first significant maximum value within a reasonable delay range. .based on A cosine-matched template is constructed and cross-correlation filtering is performed with a one-dimensional signal. After subtracting the low-frequency baseline, a periodic-based filter is applied. Minimum spacing constraints (such as) Peak finding is performed to obtain the initial set of gap peak values.

[0024] Step S6: Circular Closed-Loop Intelligent Leakage Completion Traverses the angular spacing between adjacent peaks in the initial peak set. Calculate the spacing and optimal period. ratio .like If so, it is determined that there is a missing grain at that location, and the quantity is [number missing]. Then, within that interval, according to The step size is used for equally spaced linear interpolation to fill in the missing virtual peaks.

[0025] Step S7: Biological Even-Number Row Constraint and Yield Measurement Parameter Calculation. The development of maize inflorescence division determines that the number of rows in normal maize must be even, and the number of kernels per section is equivalent to the "number of rows per ear," which is the core indicator for yield measurement per plant. The system statistically analyzes the total peak number (number of rows) after filling in the gaps and performs a modulo 2 judgment. If it is odd, it is corrected by adding or subtracting one to make it even, thus obtaining the final number of rows per ear. Finally, use proportions. Calculate the actual physical parameters: Actual physical outer diameter: (6) Actual physical inner diameter: (7) Physical length of corn kernels: (8) Theoretical physical particle width: (9) Example 2 The present invention also relates to an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method described above.

[0026] The electronic device can be a desktop computer, laptop, handheld computer, or cloud server, etc. The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The memory can be used to store computer programs and / or modules. The processor performs various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory, and by accessing data stored in the memory.

[0027] Example 3 The present invention also relates to a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described above.

[0028] Specifically, the memory may include high-speed random access memory, as well as non-volatile memory, such as hard disks, RAM, plug-in hard disks, smart media cards (SMC), secure digital (SD) cards, flash cards, at least one disk storage device, flash memory device, or other volatile solid-state storage devices.

[0029] The technical features of the embodiments described above can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification. It should be noted that the terms "in one embodiment," "for example," and "again" in this invention are intended to illustrate the invention and are not intended to limit the invention.

[0030] The embodiments described above are merely examples of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention.

Claims

1. An automatic detection method for maize cross-sectional phenotypic parameters based on image processing and polar coordinate expansion, characterized in that, Includes the following steps: S1. Acquire cross-sectional images of corn, and extract the outer contour and inner core contour of each corn cross-section in the image; at the same time, obtain scale reference information reflecting the physical scale of the shooting; S2. Pair the outer contour and inner core contour of the same corn entity, and extract the outer diameter and inner diameter that reflect the pixel size of the corn cross section; at the same time, calculate the conversion ratio from pixel to physical size according to the scale reference information. S3. Using the same corn entity as a reference, the region of interest between the outer diameter and the inner diameter is spatially transformed and expanded to obtain a rectangular expanded image band. S4. Project the rectangular unfolded image strip radially to obtain a one-dimensional signal reflecting the distribution characteristics of corn kernels or gaps between corn kernels. S5. Perform periodic analysis on the one-dimensional signal to obtain the optimal period, and obtain an initial set of peak values ​​based on the optimal period through peak detection. Each peak value corresponds to the position of a corn kernel or the gap between corn kernels. S6. Based on the optimal period and the annular closed characteristic of the corn cross section, interpolation is performed to fill any gaps that may be missed in the initial peak set. S7. Apply a biological even number constraint to the total number of peak values ​​after the leak filling treatment to determine the final number of corn kernels in the cross section. The number of kernels in the cross section represents the number of rows in the corn ear and is used to assist in the evaluation of corn yield. Combined with the conversion ratio, the outer diameter and inner diameter are converted into actual physical dimensions, and then the phenotypic parameters of the corn kernels are calculated.

2. The automatic detection method for maize cross-sectional phenotypic parameters based on image processing and polar coordinate expansion as described in claim 1, characterized in that, The scale reference information is a reference object with known physical dimensions placed on the same plane as the corn cross-section; the pixel contour of the reference object is extracted using the YOLO object detection model and the inner and outer diameters of the pixels are calculated, and the ratio of the known physical diameter of the reference object to its pixel diameter is used as the conversion ratio.

3. The automatic detection method for maize cross-sectional phenotypic parameters based on image processing and polar coordinate expansion as described in claim 1, characterized in that, The scale reference information is either depth data obtained by the depth sensor built into the shooting device, or optical imaging parameters obtained based on the camera focal length and shooting distance, and the conversion ratio is directly calculated through an optical geometric model.

4. The automatic detection method for maize cross-sectional phenotypic parameters based on image processing and polar coordinate expansion as described in claim 1, characterized in that, In step S3, the specific method of spatial transformation unfolding includes: evaluating the shape features of the paired outer contours; if the difference in the length and width ratio of the successfully paired boxes is greater than a preset deformation threshold, then the corn entity is regarded as an ellipse, and the region of interest is unfolded by ellipse-polar coordinate transformation; otherwise, it is regarded as a standard circle, with the center point of the successfully paired boxes as the origin of polar coordinates, and the outer diameter as the maximum unfolding radius, a standard polar coordinate to rectangular coordinate transformation is performed, and the annular region between the inner diameter and the outer diameter is extracted as the region of interest in the unfolded image.

5. The automatic detection method for maize cross-sectional phenotypic parameters based on image processing and polar coordinate expansion as described in claim 1, characterized in that, In step S4, the method for obtaining the one-dimensional signal is as follows: after performing grayscale conversion and image enhancement and denoising processing on the rectangular unfolded image band, the mathematical feature value of the grayscale of each column of pixels is calculated radially within a preset width segment to obtain the one-dimensional signal after dimensionality reduction, so that the gaps between corn kernels or the center of corn kernels are presented in the form of peaks in the signal.

6. The automatic detection method for maize cross-sectional phenotypic parameters based on image processing and polar coordinate expansion as described in claim 1, characterized in that, In step S5, the peak detection process includes: constructing a matching template signal based on the acquired optimal period, performing cross-correlation matched filtering operation on the matching template signal and the one-dimensional signal; performing local baseline subtraction on the filtered output signal, and then performing peak detection with minimum spacing and a preset threshold to obtain the initial peak set.

7. The automatic detection method for maize cross-sectional phenotypic parameters based on image processing and polar coordinate expansion as described in claim 1, characterized in that, In step S6, the specific method of the interpolation and omission processing is as follows: calculate the angular spacing between adjacent peak pairs, including considering the connection spacing across the ring boundary between the last peak and the first peak; when any adjacent spacing exceeds a preset multiple of the optimal period, it is determined that there is a missed detection in the gap, and the missing peak is supplemented by interpolation at equal intervals in the gap.

8. The automatic detection method for maize cross-sectional phenotypic parameters based on image processing and polar coordinate expansion as described in claim 1, characterized in that, In step S7, the specific method for applying the biological even number constraint is as follows: determine whether the total number of peak values ​​after the leak filling treatment is even; if it is odd, make corrections to meet the even number constraint; the physical length is calculated by subtracting the physical size of the inner diameter from the physical size of the outer diameter; the physical width is calculated by dividing the average circumference of the outer diameter physical size and the inner diameter physical size by the final number of corn ear rows.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the automatic detection method for corn cross-sectional phenotypic parameters based on image processing and polar coordinate expansion as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the automatic detection method for corn cross-sectional phenotypic parameters based on image processing and polar coordinate expansion as described in any one of claims 1 to 8.

Images