Method for identifying fruit tree identity by fusing tree core coordinates and local tree group visual features, terminal and storage medium
By fusing tree center coordinates with local tree group visual features, the problems of positioning error and visual feature instability in fruit tree identification were solved, achieving highly accurate and reliable fruit tree identification.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHENZHEN FENGNONG SHUZHI AGRI TECH CO LTD
- Filing Date
- 2026-03-15
- Publication Date
- 2026-06-09
AI Technical Summary
In existing technologies, the spatial positioning information of fruit trees is easily affected by occlusion and environmental changes, leading to drift in positioning results. Relying solely on visual recognition carries the risk of misjudgment, especially in densely planted orchards where the visual characteristics of individual trees are unstable and difficult to use as reliable identification markers.
A fruit tree identification method that integrates tree center coordinates and local tree group visual features determines the digital identity of fruit trees by acquiring image data within the target area, extracting relative positions and visual features, and comparing them with historical features.
It improves the accuracy and practicality of fruit tree identification, overcomes the shortcomings of relying solely on coordinates which are susceptible to errors or visual features which are susceptible to interference, adapts to changes in fruit tree growth and different phenological stages, and provides a stable identification mechanism.
Smart Images
Figure CN122176516A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of agricultural digital management technology, and in particular to a method, terminal and storage medium for fruit tree identification that integrates tree center coordinates and local tree group visual features. Background Technology
[0002] In digital management and agricultural traceability systems for fruit trees, a unique digital identity is typically assigned to each tree to achieve continuous data linking throughout its entire lifecycle. Current technologies often rely on high-precision positioning information as the primary basis for identifying fruit trees. However, spatial positioning information is susceptible to satellite signal obstruction, environmental changes, and equipment differences, leading to drift in positioning results. Relying solely on visual recognition also carries the risk of misjudgment due to tree growth changes, pruning behavior, and the high degree of similarity in appearance among trees of the same variety. Especially in densely planted orchards, the visual characteristics of individual trees are prone to repetition or instability, making them unreliable as identification markers.
[0003] Therefore, there is an urgent need for a method that can comprehensively utilize the spatial information of the fruit tree core and the visual feature information of the local fruit tree group to reliably confirm the digital identity of fruit trees. Summary of the Invention
[0004] To address the aforementioned problems, this invention proposes a fruit tree identification method that integrates tree center coordinates with local tree group visual features. Based on this method, this invention also discloses a terminal and storage medium that apply the method.
[0005] This invention is achieved through the following technical solution: A method for identifying fruit trees by fusing tree center coordinates with local tree group visual features includes: The target range is obtained by using the geographic coordinates of the tree core of the fruit tree to be confirmed and a preset spatial threshold, and then the candidate digital identities in the database are obtained by filtering. Acquire image data of local tree groups within the target area and extract the relative features of the fruit tree to be identified and other fruit trees within the target area; The relative features are compared with the historical features corresponding to each candidate digital identity, and the digital identity of the fruit tree to be confirmed is determined based on the comparison results. The relative features include: relative positional features and relative visual features.
[0006] Furthermore, acquiring image data of the local tree group within the target area specifically includes: Obtain a top-down view of a local tree group within the target area; Preprocess the top view image; The preprocessing includes at least one of the following: image distortion correction, brightness and color normalization, scale unification, image cropping, and noise suppression.
[0007] Furthermore, the relative positional features include: the relative distance and / or relative azimuth angle between the fruit tree to be identified and other fruit trees within the target range.
[0008] Furthermore, the relative visual features include: the difference in canopy morphology between the fruit tree to be identified and other fruit trees within the target range.
[0009] Furthermore, the differences in crown morphology include at least one of the following: differences in crown size, crown shape, crown density, and crown outline.
[0010] Furthermore, comparing the relative features with the historical features corresponding to each candidate digital identity, and determining the digital identity of the fruit tree to be confirmed based on the comparison results specifically includes: Calculate the similarity between the relative features and the historical features corresponding to each candidate digital identity; If there are candidate digital identities with a similarity exceeding a preset threshold, the current candidate digital identity is determined as the digital identity of the fruit tree to be confirmed. If no candidate digital identity exists with a similarity exceeding the preset threshold, a new digital identity is generated for the fruit tree to be confirmed.
[0011] Furthermore, calculating the similarity between the relative features and the historical features corresponding to each candidate digital identity specifically includes: The relative features and the historical features are represented as feature vectors; The similarity is obtained by calculating the cosine similarity, Euclidean distance, or weighted distance between the feature vectors.
[0012] Furthermore, after comparing the relative features with the historical features corresponding to each candidate digital identity, and determining the digital identity of the fruit tree to be confirmed based on the comparison results, the process further includes: Obtain the digital identity of the confirmed fruit tree, and add the current relative features to the historical feature set corresponding to the digital identity.
[0013] A storage device stores a fruit tree identification program that integrates tree center coordinates and local tree group visual features. When the fruit tree identification program is executed by a processor, it implements the steps of the fruit tree identification method that integrates tree center coordinates and local tree group visual features as described above.
[0014] A terminal includes: a memory, a processor, and a fruit tree identification program that integrates tree center coordinates and local tree group visual features, stored in the memory and executable on the processor; when the fruit tree identification program integrating tree center coordinates and local tree group visual features is executed by the processor, it implements the steps of the fruit tree identification method integrating tree center coordinates and local tree group visual features as described above.
[0015] The beneficial effects of this invention are as follows: This invention improves the accuracy and practicality of fruit tree identification by integrating spatial filtering based on tree center coordinates with visual feature matching of local tree groups. Through a dual verification mechanism of relative coordinates and visual comparison, it effectively overcomes the shortcomings of relying solely on coordinates, which is susceptible to positioning errors, or relying solely on visual features, which is susceptible to interference from light, season, and growth changes. The method utilizes the relative relationships between fruit trees as identification features, reducing reliance on the constancy of individual tree characteristics and adapting to changes in fruit tree growth and different phenological stages. Attached Figure Description
[0016] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0017] Figure 1 This is a schematic diagram of the workflow of the fruit tree identification method that integrates tree center coordinates and local tree group visual features disclosed in this invention. Figure 2 This is a schematic diagram of the internal operating environment of the terminal disclosed in this invention. Detailed Implementation
[0018] To make the technical problems, technical solutions, and beneficial effects to be solved by this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and are not intended to limit the scope of this application.
[0019] It should be understood that the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.
[0020] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.
[0021] Please refer to Figure 1 The present invention discloses a method for fruit tree identification that integrates tree center coordinates and local tree group visual features, comprising: S1. Obtain the target range based on the geographic coordinates of the tree core of the fruit tree to be confirmed and the preset spatial threshold, and filter to obtain candidate digital identities from the database. S2, acquire image data of local tree groups within the target area, and extract the relative features of the fruit tree to be identified and other fruit trees within the target area; S3, compare the relative features with the historical features corresponding to each candidate digital identity, and determine the digital identity of the fruit tree to be confirmed based on the comparison results.
[0022] The relative features include: relative position features and relative visual features.
[0023] In step S1, the geographic coordinates of the tree core of the fruit tree to be confirmed are first obtained. These coordinates determine the specific location of the fruit tree within the orchard and can be expressed as latitude and longitude coordinates or local custom coordinates within the orchard. After determining the location, a spatial threshold is set. Using the geographic coordinates of the tree core of the fruit tree to be confirmed as the center and the preset spatial threshold as the expansion range, a target area is delineated. This target area contains a local tree group, which includes the fruit tree to be confirmed and at least one adjacent fruit tree. Simultaneously with obtaining the target area, digital identities are retrieved from the database. Each digital identity corresponds to a coordinate value. Digital identities whose coordinate values fall within the target area are filtered and listed as candidate digital identities.
[0024] In one feasible embodiment of the present invention, the preset spatial threshold is a certain value between 0.2 and 0.5 meters, and the target range is a circular area with the geographic coordinates of a tree center as the center and the preset spatial threshold as the radius; four fruit trees, including the fruit tree to be confirmed, are obtained within the target range. All candidate digital identities located within the target range are retrieved from the database.
[0025] That is, step S1 is to determine the target range, and the image information and candidate digital identities of the local tree groups within the target range.
[0026] Step S2 involves image acquisition and feature extraction for the local tree group: S21, Obtain a top-down view of the local tree group within the target area; S22, Preprocess the top view image.
[0027] A drone is used to acquire top-down images of the target area, which are then preprocessed. Because the acquired images may vary in size at the same location due to different shooting angles, distortion, or local brightness differences caused by lighting conditions, preprocessing is necessary. This preprocessing includes at least one of the following: image distortion correction, brightness and color normalization, scale unification, image cropping, and noise suppression. Preprocessing ensures that the acquired images accurately reflect the local tree clusters within the target area.
[0028] A top-down image of a local tree group is used to extract relative features, which are obtained by comparing the fruit tree to be identified with other fruit trees within the target area. Specifically: Relative positional characteristics: The relative distance and / or relative azimuth of the fruit tree to be identified within the target area compared to other fruit trees. For example, if there are two other trees within the target area, one located 1.2 meters due south of the fruit tree to be identified, and the other located 2 meters 45° east of north of the fruit tree to be identified, the relative positional characteristics of the fruit tree to be identified can be obtained based on information about at least one other fruit tree within the target area. Relative positional characteristics determine the location of the fruit tree through its spatial distribution within the target area, avoiding inaccurate judgments due to inaccurate positioning during photography. Relative visual features: The difference in crown morphology between the fruit tree to be identified and other fruit trees within the target range. Due to the influence of fruit tree growth, there is a great deal of uncertainty in comparing a single fruit tree before and after. However, the target fruit tree can be identified by comparing the crown shapes of historically adjacent fruit trees: for example, in historical images, the crown size of fruit tree A is larger than that of fruit tree B, but smaller than that of fruit tree C; in the most recently acquired historical images, within the same target range, there are three fruit trees with different crown sizes, so the fruit tree with the middle crown size can be identified as fruit tree A. In one feasible embodiment of the present invention, the crown morphology difference includes at least one of the following: crown size difference (crown coverage area when viewed from above), crown shape difference (crown is round or elliptical, elliptic eccentricity), crown density difference (crown branch and leaf density), and crown outline difference (crown boundary regularity, number of gaps in the top-view outline).
[0029] Based on the above, step S2 specifically includes: S23, calculate the similarity between the relative features and the historical features corresponding to each candidate digital identity; S24, if there are candidate digital identities with a similarity exceeding a preset threshold, then the current candidate digital identity is determined as the digital identity of the fruit tree to be confirmed. S25, if there are no candidate digital identities with a similarity exceeding the preset threshold, then a new digital identity is generated for the fruit tree to be confirmed.
[0030] The relative positional and visual features of the fruit trees to be identified within the target area are transformed into feature vectors. The historical features corresponding to the candidate digital identities are compared with these feature vectors. The historical features are the feature vectors transformed from the relative positional and visual features of the corresponding fruit trees during previous identity recognition processes. In other words, the currently acquired feature vectors are compared with the historical feature vectors. The similarity of the feature vectors is calculated using one of cosine similarity, Euclidean distance, or weighted distance.
[0031] If the similarity exceeds the preset threshold, it means that the fruit tree to be confirmed meets one of the historical features, and the digital identity of the fruit tree is confirmed to be the digital identity corresponding to the historical feature.
[0032] If the similarity scores do not exceed the threshold, it means that the fruit tree to be confirmed is a newly appeared fruit tree in the orchard. At this time, a new digital identity is generated and assigned to the fruit tree.
[0033] Whether determining whether a fruit tree has an existing digital identity or a new digital identity, it is necessary to associate the relative characteristics with the digital identity. Therefore, after step S25, the following steps are also included: S26, obtain the digital identity of the confirmed fruit tree, and add the current relative feature to the historical feature set corresponding to the digital identity.
[0034] Based on the above method, the present invention also discloses a storage device, wherein the storage device stores a fruit tree identification program that integrates tree center coordinates and local tree group visual features. When the fruit tree identification program that integrates tree center coordinates and local tree group visual features is executed by a processor, it implements the steps of the fruit tree identification method that integrates tree center coordinates and local tree group visual features as described above.
[0035] Please refer to Figure 2 Based on the above method, the present invention also discloses a terminal, comprising: a memory, a processor, and a fruit tree identification program that integrates tree center coordinates and local tree group visual features, stored in the memory and executable on the processor; when the fruit tree identification program that integrates tree center coordinates and local tree group visual features is executed by the processor, it implements the steps of the fruit tree identification method that integrates tree center coordinates and local tree group visual features as described above.
[0036] This invention discloses a fruit tree identification method that integrates tree center coordinates and local tree group visual features, significantly improving the accuracy of identification: it combines spatial filtering using tree center coordinates with matching visual features of local tree groups to form a dual verification mechanism. First, spatial thresholding quickly identifies candidate targets, eliminating a large number of irrelevant individuals; then, it uses the relative features of the local tree group centered on the target tree for precise comparison. The overall strategy of "coarse selection followed by precise matching" overcomes the shortcomings of relying solely on coordinates, which are easily affected by positioning errors, or relying solely on visual features, which are easily affected by changes in light, season, and fruit tree growth. This achieves high accuracy and high reliability in complex orchard environments.
[0037] This invention effectively utilizes the stable positional and visual relationships among fruit trees, enhancing the uniqueness and stability of features: It extracts not isolated features of a single fruit tree, but rather the relative position and visual characteristics between the fruit tree to be identified and its surrounding trees. While the absolute appearance of a fruit tree changes with growth, pruning, and season, its local comparative relationships (such as relative distance, orientation, and canopy morphology comparison) change little. This feature, based on the local tree group structure, possesses stronger uniqueness and time-invariance, providing a stable and difficult-to-replicate "relationship fingerprint" for the fruit tree, greatly improving the long-term effectiveness of identification. It reduces reliance on the precision and constancy of individual tree features, adapting to changes in appearance caused by natural growth, artificial pruning, and different phenological stages, thus having broader applicability and lifecycle coverage.
[0038] This invention establishes a traceable and updatable digital file for each fruit tree. This provides basic data support for achieving precision agricultural operations (such as fertilization, spraying, and harvesting) based on individual trees, yield tracking, growth model construction, and full-chain quality traceability, enabling refined management and monitoring of orchards.
[0039] This invention requires minimal computation, has high accuracy, and can improve the overall operating efficiency of the recognition system, demonstrating good practicality and deployability.
[0040] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the protection scope of this application.
Claims
1. A method for fruit tree identification that integrates tree center coordinates and local tree group visual features, characterized in that, include: The target range is obtained by using the geographic coordinates of the tree core of the fruit tree to be confirmed and a preset spatial threshold, and then the candidate digital identities in the database are obtained by filtering. Acquire image data of local tree groups within the target area and extract the relative features of the fruit tree to be identified and other fruit trees within the target area; The relative features are compared with the historical features corresponding to each candidate digital identity, and the digital identity of the fruit tree to be confirmed is determined based on the comparison results. The relative features include: relative positional features and relative visual features.
2. The method according to claim 1, characterized in that, The acquisition of image data of the local tree group within the target range specifically includes: Obtain a top-down view of a local tree group within the target area; Preprocess the top view image; The preprocessing includes at least one of the following: image distortion correction, brightness and color normalization, scale unification, image cropping, and noise suppression.
3. The method according to claim 1, characterized in that, The relative position features include: the relative distance and / or relative azimuth angle between the fruit tree to be identified and other fruit trees within the target range.
4. The method according to claim 1, characterized in that, The relative visual features include: the difference in canopy morphology between the fruit tree to be identified and other fruit trees within the target range.
5. The method according to claim 4, characterized in that, The differences in tree crown morphology include at least one of the following: differences in tree crown size, differences in tree crown shape, differences in tree crown density, and differences in tree crown outline.
6. The method according to claim 1, characterized in that, The process of comparing the relative features with the historical features corresponding to each candidate digital identity, and determining the digital identity of the fruit tree to be confirmed based on the comparison results, specifically includes: Calculate the similarity between the relative features and the historical features corresponding to each candidate digital identity; If there are candidate digital identities with a similarity exceeding a preset threshold, the current candidate digital identity is determined as the digital identity of the fruit tree to be confirmed. If no candidate digital identity exists with a similarity exceeding the preset threshold, a new digital identity is generated for the fruit tree to be confirmed.
7. The method according to claim 6, characterized in that, Calculating the similarity between the relative features and the historical features corresponding to each candidate digital identity specifically includes: The relative features and the historical features are represented as feature vectors; The similarity is obtained by calculating the cosine similarity, Euclidean distance, or weighted distance between the feature vectors.
8. The method according to claim 1 or 6, characterized in that, After comparing the relative features with the historical features corresponding to each candidate digital identity, and determining the digital identity of the fruit tree to be confirmed based on the comparison results, the process further includes: Obtain the digital identity of the confirmed fruit tree, and add the current relative features to the historical feature set corresponding to the digital identity.
9. A storage device, characterized in that, The storage device stores a fruit tree identification program that integrates tree center coordinates and local tree group visual features. When the fruit tree identification program that integrates tree center coordinates and local tree group visual features is executed by the processor, it implements the steps of the fruit tree identification method that integrates tree center coordinates and local tree group visual features as described in any one of claims 1 to 8.
10. A terminal, characterized in that, The terminal includes: a memory, a processor, and a fruit tree identification program that integrates tree center coordinates and local tree group visual features, stored in the memory and executable on the processor; when the fruit tree identification program that integrates tree center coordinates and local tree group visual features is executed by the processor, it implements the steps of the fruit tree identification method that integrates tree center coordinates and local tree group visual features as described in any one of claims 1 to 8.