CN110653166A - Fruit detection and classification method and device - Google Patents
Fruit detection and classification method and device Download PDFInfo
- Publication number
- CN110653166A CN110653166A CN201910951413.0A CN201910951413A CN110653166A CN 110653166 A CN110653166 A CN 110653166A CN 201910951413 A CN201910951413 A CN 201910951413A CN 110653166 A CN110653166 A CN 110653166A
- Authority
- CN
- China
- Prior art keywords
- fruit
- dimensional image
- images
- volume
- fruits
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
- B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
- B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
- B07C5/04—Sorting according to size
- B07C5/10—Sorting according to size measured by light-responsive means
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
- B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
- B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
- B07C5/34—Sorting according to other particular properties
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
- B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
- B07C2501/00—Sorting according to a characteristic or feature of the articles or material to be sorted
- B07C2501/009—Sorting of fruit
Landscapes
- Sorting Of Articles (AREA)
Abstract
The invention relates to a fruit detection grading method and a device, belonging to the field of fruit and vegetable detection, wherein the method comprises the steps of collecting a plurality of images of each fruit, extracting characteristic points in each image, matching the characteristic points of all the images, and performing three-dimensional reconstruction according to the matched characteristic points to obtain a three-dimensional image of each fruit; dividing a target area and a background area in the three-dimensional image of each fruit, slicing the three-dimensional image of each fruit, and calculating to obtain the volume of each fruit according to the area of the target area in each layer of slices; and classifying the fruits according to the volume of the obtained fruits. Compared with the prior art, the fruit grading method based on the fruit diameter can effectively grade the fruits according to the fruit volume grading method, and can calculate the fruit volume for the fruits with complex and irregular shapes, so that the fruit grading reliability is improved.
Description
Technical Field
The invention belongs to the field of fruit and vegetable detection, and particularly relates to a fruit detection grading method and device.
Background
The traditional fruit and vegetable grading method is that the fruits and vegetables are graded in a manual sorting mode, time and labor are wasted, and the problems of inconsistent and inaccurate judgment of different people exist, so that the accurate sorting of the fruits becomes the primary factor for improving domestic and foreign markets, and therefore a relatively objective judgment standard is required to automatically classify the fruits. It is common practice at present to grade fruit weight using an electronic scale or fruit size using a mechanical method, and thus the grading accuracy is poor and the efficiency is low.
The existing fruit size grading method based on computer vision adopts a conventional digital image processing algorithm, calculates the fruit diameter by preprocessing a collected fruit image, segmenting a fruit region, detecting characteristics and the like, determines the actual measured value of the fruit by using the fruit diameter as the characteristic parameter of the fruit size through system calibration, and finally realizes the fruit size grading through the measured value. The method has the defects that the reliability of detection is not high, certain deviation exists in the grading result of the sizes of fruits with complex and irregular shapes, and the practical popularization and application of the method in the field of agricultural production are limited to a certain extent.
Disclosure of Invention
The invention aims to provide a fruit detection and classification method and device, which are used for solving the problem that the fruit classification method in the prior art is low in reliability and cannot realize effective classification of fruits.
Based on the purposes, the technical scheme of the fruit detection and classification method is as follows:
acquiring a plurality of images of each fruit, wherein the plurality of images of each fruit are images acquired from at least two different angles;
extracting characteristic points in each image, matching the characteristic points of all the images, and performing three-dimensional reconstruction according to the matched characteristic points to obtain a three-dimensional image of each fruit;
dividing a target area and a background area in the three-dimensional image of each fruit, slicing the three-dimensional image of each fruit, and calculating to obtain the volume of each fruit according to the area of the target area in each layer of slices;
and classifying the fruits according to the volume of the obtained fruits.
Based on above-mentioned purpose, a fruit detects grading plant's technical scheme as follows:
comprising a memory and a processor, and a computer program stored on the memory and running on the processor, the processor being coupled to the memory, the processor implementing the above fruit detection grading method when executing the computer program.
The two technical schemes have the beneficial effects that:
according to the invention, the images of the fruits are acquired from different angles, the characteristic points of the images are extracted and matched, the matched characteristic points are utilized to carry out three-dimensional reconstruction, the three-dimensional images of the fruits can be obtained, slicing processing is carried out, then the volume of the fruits can be calculated according to the area of the target area in each layer of slices, and finally the fruits are graded according to the volume of the fruits. Compared with the prior art, the fruit grading method based on the fruit diameter can effectively grade the fruits according to the fruit volume grading method, and can calculate the fruit volume for the fruits with complex and irregular shapes, so that the fruit grading reliability is improved.
In order to improve the reliability of fruit grading, furthermore, the fruits are graded according to the density of the fruits, wherein the density of the fruits is calculated according to the volume and the weight of the fruits, the moisture of the fruits can be represented through the volume and the density of the fruits, the hardness of the fruits can be represented through the density, and equivalently, the moisture and the hardness of the fruits are considered at the same time of fruit grading, so that the fine grading of the fruits is realized, and the grading reliability is improved.
Further, the fruits are also graded in combination with their surface area, which is obtained by the steps of:
and obtaining contour points of a target area in the three-dimensional image of each fruit, performing contour smoothing on the basis of the contour points, establishing a triangular patch, and calculating the area of a curved surface, thereby obtaining the surface area of the apple.
In the fruit classification, due to the surface area of the added fruit, when the fruit is classified by using the volume and the surface area of the fruit in combination, the fruit can be further finely classified, and the fruit with higher quality can be preferably selected, for example, the fruit with higher moisture content and relatively smaller surface area is selected. When the fruit is classified by a combination of its volume, density and surface area, the fruit can be optimally classified, for example, fruits having a high moisture content, a relatively small surface area and a high hardness can be selected.
In order to improve the precision of three-dimensional reconstruction, furthermore, for a plurality of images of each fruit, the fruits are acquired from at least two different angles by taking the whole surface of the covered fruit as a target, and an accurate fruit three-dimensional image can be established.
Further, matching the feature points of all the images includes:
sparse matching and sparse reconstruction are carried out on the extracted feature points by utilizing a rapid multi-view point feature matching method of a minimum spanning tree to obtain sparse point cloud;
and taking the sparse point cloud as a reference point, and performing dense matching and dense reconstruction on the obtained feature points by using a multi-view dense matching method to obtain a data model of the fruit three-dimensional image.
The data model of the fruit three-dimensional image obtained by the two methods is more accurate, and the accuracy of the fruit three-dimensional image is improved.
Further, the performing three-dimensional reconstruction includes:
and performing surface reconstruction on the data model of the fruit three-dimensional image by using a point-by-point insertion algorithm to obtain the fruit three-dimensional image so as to ensure the imaging quality of the three-dimensional image.
Drawings
FIG. 1 is a flow chart of a method of feature point extraction and matching of the present invention;
FIG. 2 is a flow chart of the clustering algorithm of the present invention;
FIG. 3 is a flow chart of a three-dimensional measurement algorithm of the present invention;
FIG. 4 is a schematic flow chart of the fruit detection and classification method of the present invention.
Detailed Description
The following further describes embodiments of the present invention with reference to the drawings.
Example 1:
the invention provides a fruit detection grading device, which comprises an image acquisition device and a processor for acquiring and connecting the image acquisition device, wherein optionally, the image acquisition device comprises three CCD cameras which are arranged at different positions of a grading production line and can acquire articles passing through the grading production line from different angles; the processor is a computer terminal and is used for receiving and processing images sent by the three CCD cameras.
Based on the device, the fruit detection and classification method is introduced by taking apple classification as an example, and specifically comprises the following steps:
step one, in the transmission process of apples on a grading production line, the apples are lifted by a fruit cup and rotate, a certain distance is reserved between every two adjacent apples on the grading production line, the apples sequentially enter an image acquisition range area of a CCD camera according to the front and back sequence, the three CCD cameras acquire images on the surfaces of the apples, the image acquisition at different angles covers the whole surfaces of the apples, therefore, one apple corresponds to a plurality of images, and finally the acquired images of the apples are transmitted to a computer terminal for storage.
And secondly, the computer terminal performs feature point extraction, sparse matching and sparse reconstruction, dense matching and dense reconstruction, surface reconstruction (namely three-dimensional reconstruction) and slicing processing according to the obtained image of the apple, calculates the approximate stage volume, further calculates the whole volume of the apple and grades the apple. Specifically, the method comprises the following substeps:
step 1), the flow of the feature point extraction and matching method is shown in fig. 1, and the purpose is to extract and match feature points of a plurality of images, based on the purpose, firstly, a plurality of images of apples stored in a computer terminal are obtained, the first image is used as an original image, the second image is used as a target image, a plurality of feature points are extracted from the two images by sequentially adopting a feature point positioning and describing algorithm to obtain a feature point set of a target (namely the apple to be classified), and then sparse matching and sparse reconstruction of the feature points are realized by using a fast multi-view point feature matching method based on a minimum spanning tree to obtain sparse point cloud. And then, taking the second image as an original image, taking the third image as a target image, and matching the feature points, and so on.
Specifically, the feature point positioning and describing algorithm adopted by the embodiment is an SIFT feature point detection algorithm and an SIFT feature point descriptor algorithm, and the algorithm can have certain robustness on scale change, rotation change and radiation change among images and is widely applied to stereo matching. The detection process Of the SIFT feature point detection algorithm is completed in a Gaussian Difference (DOG) scale space, a Gaussian Difference image is obtained by subtracting adjacent upper and lower layers Of images in a Gaussian pyramid, and then Hessian matrix detection and Taylor expansion are carried out, so that the accurate position Of the feature point is positioned.
In this embodiment, the SIFT feature point descriptor algorithm is a local feature descriptor, and describes the feature points and the surrounding neighborhood pixel point information. The algorithm adopts an image gradient method to estimate the stable direction of the local area of the feature point as the main direction of the feature point according to the local feature of the image, thereby realizing the rotational invariance of the descriptor. On the basis of image gradient calculation, a feature point neighborhood window is divided into a plurality of regions with the same size, and statistical results of gradients of the regions in 8 directions are calculated respectively. In order to improve the calculation accuracy, the rotation main directions of the feature points are classified into a histogram including eight directions after being weighted by Gaussian.
In this embodiment, the method for fast matching features of multi-view points based on a minimum spanning tree is implemented by calculating the euclidean distance between 128-dimensional key points of two sets of feature points, and when the euclidean distance between feature vectors corresponding to two feature points is smaller than a set threshold, the two feature points are considered to correspond to the same scene point, and a sparse point cloud is formed after the feature points are matched.
Step 2), the dense matching and dense reconstruction algorithm adopted by the invention is a multi-view dense matching (CMVS/PMVS) method, the method uses CMVS (clustering multi-view stereo) algorithm to classify and optimize SFM (structure from motion) input for image clustering, uses sparse point cloud as a reference point, reduces dense matching time and space cost, and the main realization steps are as follows: the method comprises the steps of firstly, fusing SFM points, removing redundant images, enhancing cluster size constraint and increasing image coverage constraint. Thereby acquiring a plurality of image clusters, and the flow is shown in fig. 2.
And then, performing matching, expanding and filtering through a pulse-based multi-view stereo (PMVS) algorithm to complete dense matching, wherein the PMVS algorithm is based on dense matching of a patch model, and reconstructing each image cluster acquired by the CMVS algorithm to obtain dense point cloud (dense reconstruction) and a data model of the apple three-dimensional image (data model formed by the dense point cloud).
And 3) performing filtering smoothing treatment on the dense point cloud by using a point cloud smoothing algorithm, wherein the point cloud smoothing algorithm used in the embodiment is a Bilateral filtering (Bilateral filtering) algorithm which is an improvement of Gaussian filtering, and when the weight occupied by the neighborhood pixels in the convolution process is calculated, the geometric proximity degree of the neighborhood pixels to the current central pixel and the similarity degree of the gray value are considered, and the weight is determined by the combined action of the neighborhood pixels and the gray value, so that the edge characteristic information of the image can be retained while the image is filtered and smoothed.
And 4), the surface reconstruction algorithm adopted by the invention is a general algorithm of Delaunay triangulation, namely a point-by-point insertion algorithm, the algorithm is strict in theory and good in uniqueness based on a scattered point network construction algorithm, a grid meets the empty circle characteristic, the algorithm is ideal, and after network construction is finished and new points are added, the network construction is not required to be carried out on all the points, and only the influence triangle range of the new points is required to be locally networked. The method mainly comprises the following steps:
s1, constructing a super triangle, including all scattered points (dense point clouds), and placing the super triangle into a triangle linked list;
s2, sequentially inserting scattered points in the point set, finding out a triangle of which an external circle comprises an insertion point from the triangular linked list, deleting a common edge of the influencing triangle, and connecting the insertion point with all vertexes of the influencing triangle so as to complete the insertion of one point in the Delaunay triangular linked list;
s3, optimizing the local newly formed triangles according to an optimization criterion, and putting the formed triangles into a Delaunay triangle linked list;
and S4, circularly executing the step S2 until all scattered points are inserted completely, thereby realizing surface reconstruction, namely three-dimensional reconstruction and obtaining a three-dimensional image of the apple.
And 5), the calculation process of the whole volume of the apple adopted by the invention is shown in fig. 3, the three-dimensional image of the apple is segmented through a C-V model to obtain a target area and a background area, the three-dimensional image of the fruit is sliced, the area of the target area of each layer of slices is calculated, and further the measurement of the volume is realized.
Specifically, the area of each slice target area is calculated by using a discretized green formula, wherein the discretized green formula is as follows:
wherein S is the area of each layer of slice target area, i is the serial number of the slice, c is the total number of the slices, xiAs the X coordinate, y, of slice iiAs Y-coordinate, x, of slice ii+1Is the X coordinate, y, of slice i +1i+1Is the Y coordinate of slice i + 1.
Approximately regarding each two layers of slices as a table body, and calculating the target volume by using a table body formula as follows:
wherein V is the volume of the table body between two layers of slices, hiIs the layer thickness between two adjacent slices, Sslice,i,Sslice,i+1The area of the target region of each i, i +1 th layer is obtained by the above discretized green's equation, and n represents the number of slice layers (the same meaning as that represented by c).
After n-1 table body volumes are obtained, accumulating all table body volumes to obtain the whole volume v of the apple, and then dividing the apple into three equal times according to the volume size of the apple, wherein the three equal times are respectively as follows: 200cm3Below, 200cm3To 300cm3And 300cm3The above. The specific number of the equal times is not limited in this embodiment, and as other embodiments, the apple can be divided into two equal times or more than three equal times according to actual requirements.
According to the invention, the images of the fruits are acquired from different angles, the characteristic points of the images are extracted and matched, the matched characteristic points are utilized to carry out three-dimensional reconstruction, the three-dimensional images of the fruits can be obtained, slicing processing is carried out, then the volume of the fruits can be calculated according to the area of the target area in each layer of slices, and finally the fruits are graded according to the volume of the fruits. Compared with the prior art, the fruit grading method based on the fruit diameter can effectively grade the fruits according to the fruit volume grading method, and can calculate the fruit volume for the fruits with complex and irregular shapes, so that the fruit grading reliability is improved.
In this embodiment, the fruits are classified only according to the volumes of the fruits, as another embodiment, other factors may be added to combine the volumes of the fruits to perform more detailed classification on the fruits, for example, contour points of a target area in a three-dimensional image of each fruit are obtained, contour smoothing is performed on the basis of the contour points, a triangular surface patch is established by using the smoothed contour points, and the surface area of the apple is calculated, thereby obtaining the surface area of the apple.
In the fruit classification, due to the surface area of the added fruit, when the fruit is classified by using the volume and the surface area of the fruit in combination, the fruit can be further finely classified, and the fruit with higher quality can be preferably selected, for example, the fruit with higher moisture content and relatively smaller surface area is selected.
Example 2:
the invention provides a fruit detecting and grading device, which comprises: the system comprises an online weighing device, an image acquisition device and a processor for acquiring and connecting the online weighing device and the image acquisition device, wherein optionally, the online weighing device is a weight sensor and is arranged at an inlet of the production line; the image acquisition devices are two CCD cameras which are arranged in different directions of the assembly line and can acquire all surfaces of articles passing through the assembly line; the processor is a computer terminal and is used for receiving and processing the apple weight data and the image data sent by the weight sensor and the CCD camera.
Based on the above device, taking apple classification as an example, the fruit detection and classification method of the present invention is introduced, as shown in fig. 4, and specifically includes the following steps:
at first let treat hierarchical apple get into the assembly line, obtain the weight information of apple through online weighing device at assembly line entrance, transmit apple weight m for computer terminal and store.
Then, the volume of each apple is calculated according to the contents of the first step and the second step in the above embodiment 1, and finally, the obtained apple weight m and the apple volume information v are used according to the formulaObtaining a density rho of the apple from the appleAnd judging the hardness of the apples according to the fruit density, and continuously grading the hardness of the apples according to the hardness.
Compared with the prior art that the size of the fruits is graded, the method disclosed by the invention grades the quality of the apples according to the weight, the volume and the density of the apples, realizes intelligent grading of picked high-quality fruits and vegetables, and carries out multi-channel real-time online sorting on one or more apples, so that the sorting precision and efficiency of the apples are improved, and the industrial requirements are met.
In addition, the processor in this embodiment may be a computer terminal, or may be a microprocessor, such as an ARM, or may be a programmable chip, such as an FPGA, a DSP, or the like, and the hardware configuration of the processor is not limited in this embodiment, but a computer program executed in the processor can implement the fruit detection and classification method in embodiment 1 or embodiment 2.
The above description is only two preferred embodiments of the present invention, and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. For example, in the present embodiment, an apple is classified, and as another embodiment, the classification method of the present invention is also applicable to other fruits, such as oranges, grapefruits, muskmelons, kiwis, and the like, and thus the classification object of the present invention is not limited to apples.
As another example, in this embodiment, the image acquisition from at least two different angles covers the whole surface of a certain fruit, so that an accurate three-dimensional image of the fruit can be established. In order to acquire apple images from different angles, one embodiment is to arrange cameras from different directions to shoot the apples on the grading production line, and the other embodiment is to adopt the fruit cup on the production line to lift and rotate, namely the apples on the fruit cup rotate to shoot and image the apples. In addition, if the acquired image does not completely cover the whole surface of the fruit to be graded, if the uncovered part is very small, the three-dimensional reconstruction of the characteristic points is not influenced, and the three-dimensional image of the fruit can still be obtained.
As another possible implementation, when the image acquisitions from at least two different angles do not cover the whole surface of a fruit, the data processing method may be adopted to fill up the missing apple surface, so as to implement three-dimensional reconstruction of the image set of each fruit.
In another example, the fruit can be classified by a combination of volume, density and surface area of the fruit, which can be optimized for selecting fruit with high moisture content, relatively small surface area and high hardness.
Claims (7)
1. A fruit detection and classification method is characterized by comprising the following steps:
acquiring a plurality of images of each fruit, wherein the plurality of images of each fruit are images acquired from at least two different angles;
extracting characteristic points in each image, matching the characteristic points of all the images, and performing three-dimensional reconstruction according to the matched characteristic points to obtain a three-dimensional image of each fruit;
dividing a target area and a background area in the three-dimensional image of each fruit, slicing the three-dimensional image of each fruit, and calculating to obtain the volume of each fruit according to the area of the target area in each layer of slices;
the individual fruits were graded according to the volume of the fruit obtained.
2. The fruit detection grading method according to claim 1, wherein each fruit is further graded in combination with the density of the fruit calculated from the volume and weight of the fruit.
3. Fruit detection grading method according to claim 1 or 2, characterized in that each fruit is further graded in combination with its surface area, which is obtained by:
and obtaining contour points of a target area in the three-dimensional image of each fruit, performing contour smoothing on the basis of the contour points, establishing a triangular patch, and calculating the area of a curved surface, thereby obtaining the surface area of the apple.
4. The fruit detection grading method according to claim 1, wherein for each fruit several images are taken of the fruit from at least two different angles with the aim of covering the entire surface of the fruit.
5. The fruit detection grading method according to claim 1, wherein matching feature points of all images comprises:
sparse matching and sparse reconstruction are carried out on the extracted feature points by utilizing a rapid multi-view point feature matching method of a minimum spanning tree to obtain sparse point cloud;
and taking the sparse point cloud as a reference point, and performing dense matching and dense reconstruction on the obtained feature points by using a multi-view dense matching method to obtain a data model of the fruit three-dimensional image.
6. The fruit detection grading method according to claim 5, wherein said performing a three-dimensional reconstruction comprises:
and performing surface reconstruction on the data model of the fruit three-dimensional image by using a point-by-point insertion algorithm to obtain the fruit three-dimensional image.
7. A fruit detection grading device comprising a memory and a processor, and a computer program stored on the memory and run on the processor, the processor being coupled to the memory, the processor when executing the computer program implementing the fruit detection grading method of any of claims 1-6 below.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910951413.0A CN110653166B (en) | 2019-10-08 | 2019-10-08 | Fruit detection and classification method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910951413.0A CN110653166B (en) | 2019-10-08 | 2019-10-08 | Fruit detection and classification method and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110653166A true CN110653166A (en) | 2020-01-07 |
CN110653166B CN110653166B (en) | 2021-10-22 |
Family
ID=69038626
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910951413.0A Active CN110653166B (en) | 2019-10-08 | 2019-10-08 | Fruit detection and classification method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110653166B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115880354A (en) * | 2023-03-02 | 2023-03-31 | 成都工业学院 | Method for calculating crown volume based on point cloud self-adaptive slicing |
CN116817754A (en) * | 2023-08-28 | 2023-09-29 | 之江实验室 | Soybean plant phenotype extraction method and system based on sparse reconstruction |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103307978A (en) * | 2013-05-23 | 2013-09-18 | 浙江大学 | Method for measuring volume and surface area of egg |
CN105716514A (en) * | 2016-01-29 | 2016-06-29 | 上海海洋大学 | Lossless detection method for fat volume of hepatopancreas and gonad of Eriocheir sinensis |
CN107899966A (en) * | 2017-12-06 | 2018-04-13 | 罗凯缤 | The shaddock quality sorting unit of view-based access control model technology |
CN109590231A (en) * | 2018-12-19 | 2019-04-09 | 上海易持自动系统有限公司 | A kind of non-regular shape material image measurement measuring and controlling device and method |
CN110047139A (en) * | 2019-04-28 | 2019-07-23 | 南昌航空大学 | A kind of specified target three-dimensional rebuilding method and system |
-
2019
- 2019-10-08 CN CN201910951413.0A patent/CN110653166B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103307978A (en) * | 2013-05-23 | 2013-09-18 | 浙江大学 | Method for measuring volume and surface area of egg |
CN105716514A (en) * | 2016-01-29 | 2016-06-29 | 上海海洋大学 | Lossless detection method for fat volume of hepatopancreas and gonad of Eriocheir sinensis |
CN107899966A (en) * | 2017-12-06 | 2018-04-13 | 罗凯缤 | The shaddock quality sorting unit of view-based access control model technology |
CN109590231A (en) * | 2018-12-19 | 2019-04-09 | 上海易持自动系统有限公司 | A kind of non-regular shape material image measurement measuring and controlling device and method |
CN110047139A (en) * | 2019-04-28 | 2019-07-23 | 南昌航空大学 | A kind of specified target three-dimensional rebuilding method and system |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115880354A (en) * | 2023-03-02 | 2023-03-31 | 成都工业学院 | Method for calculating crown volume based on point cloud self-adaptive slicing |
CN116817754A (en) * | 2023-08-28 | 2023-09-29 | 之江实验室 | Soybean plant phenotype extraction method and system based on sparse reconstruction |
CN116817754B (en) * | 2023-08-28 | 2024-01-02 | 之江实验室 | Soybean plant phenotype extraction method and system based on sparse reconstruction |
Also Published As
Publication number | Publication date |
---|---|
CN110653166B (en) | 2021-10-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111724433B (en) | Crop phenotype parameter extraction method and system based on multi-view vision | |
CN109584281B (en) | Overlapping particle layering counting method based on color image and depth image | |
CN109146948B (en) | Crop growth phenotype parameter quantification and yield correlation analysis method based on vision | |
CN112489212A (en) | Intelligent three-dimensional mapping method for building based on multi-source remote sensing data | |
CN110443836A (en) | A kind of point cloud data autoegistration method and device based on plane characteristic | |
CN109255757B (en) | Method for segmenting fruit stem region of grape bunch naturally placed by machine vision | |
CN105139379B (en) | Based on the progressive extracting method of classified and layered airborne Lidar points cloud building top surface | |
CN108171720A (en) | A kind of oblique photograph model object frontier probe method based on geometrical statistic information | |
CN107392929B (en) | Intelligent target detection and size measurement method based on human eye vision model | |
CN110653166B (en) | Fruit detection and classification method and device | |
CN107437068B (en) | Pig individual identification method based on Gabor direction histogram and pig body hair mode | |
CN110070567A (en) | A kind of ground laser point cloud method for registering | |
CN109559324A (en) | A kind of objective contour detection method in linear array images | |
CN102446356A (en) | Parallel and adaptive matching method for acquiring remote sensing images with homogeneously-distributed matched points | |
CN113470106A (en) | Non-contact cow body size information acquisition method | |
CN116721121A (en) | Plant phenotype color image feature extraction method | |
CN116523898A (en) | Tobacco phenotype character extraction method based on three-dimensional point cloud | |
CN114842262A (en) | Laser point cloud ground object automatic identification method fusing line channel orthographic images | |
CN107977968B (en) | Building layered detection method based on building shadow information mining | |
CN110942102B (en) | Probability relaxation epipolar matching method and system | |
Gong et al. | Point cloud segmentation of 3D scattered parts sampled by RealSense | |
CN115937151B (en) | Method for judging curling degree of crop leaves | |
Zhao et al. | Automatic sweet pepper detection based on point cloud images using subtractive clustering | |
Ma et al. | Rapid estimation of apple phenotypic parameters based on 3D reconstruction | |
Mohammed Amean et al. | Automatic plant branch segmentation and classification using vesselness measure |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |