CN-115249233-B - Cherry fruit dynamic growth inversion method

Abstract

The invention discloses a cherry fruit dynamic growth inversion method, which is characterized in that the growth and development conditions of fruits are monitored remotely, when cherry fruits enter a coloring period, the area of the fruits at each stage of the coloring period is obtained, the coloring days are taken as independent variables, the area is taken as dependent variables, the research on the growth trend and the growth rhythm of the cherry fruits is carried out, a fruit dynamic growth curve is fitted through the inversion method, the fruit growth rule is further given, the cherry fruit dynamic growth inversion method is finally constructed, and after a growth curve model is constructed, the reliability of the model is verified through the use of an accurate factor and a deviation factor. The method can realize rapid and accurate detection of the fruit growth state in a natural scene, dynamically grasp the growth rule of greenhouse crops, and further improve the yield and quality of cherry fruits.

Inventors

• HU LINGYAN
• LI LI
• WANG ZUMIN
• XU WEI
• LI GUOQIANG
• Qiu Shaohang
• GU MAOMAO
• SUN HAO

Assignees

• 大连大学

Dates

Publication Date

20260512

Application Date

20220414

Claims (4)

1. 1. A cherry fruit dynamic growth inversion method, which is characterized by comprising the following steps: acquiring a cherry fruit image in a coloring period; preprocessing the cherry fruit image to enable the cherry fruit image to be more in line with the original shape of the cherry fruit; obtaining the projection area of cherry fruits by improving an intelligent scissors algorithm; fitting a fruit dynamic growth curve according to an inversion method; The cherry fruit projection area is obtained by improving an intelligent scissors algorithm, and is specifically as follows: Firstly, obtaining a cherry fruit image under a natural environment obtained by opening and closing operation, detecting the edge of the image by using a Canny algorithm, finding out candidate circle centers, determining the radius according to the supporting degree of the edge of all the candidate circle centers on the cherry fruit image, obtaining the optimal radius r, extracting a target contour by using a gradient Hough transformation circle detection method, obtaining an initial edge image, and marking the obtained initial edge point as a set ; Taking an initial edge point set obtained by gradient Hough transformation circle detection as a precondition, carrying out semantic segmentation on key feature areas of cherry fruits pixel by combining a computer vision technology, adaptively searching the key feature areas of cherry fruit images, and obtaining the areas of the key feature areas, namely the projection areas of cherry fruit images in a coloring period; Fitting a fruit dynamic growth curve according to an inversion method, wherein the method specifically comprises the following steps: The projection area of cherry fruit image in coloring period is set as y, the number of days in coloring period is set as x, and a group of data pairs of x and y are formed , ) Where i=1, 2,..m, m is the total data pair number, each Are different from each other; fitting an analytical expression which is suitable for the dynamic growth rule of fruits, Reflecting the dependency between x and y, Called inversion model; The inversion model is based on data pairs , ) Obtaining a polynomial of degree n, where n m: Wherein, the A curve is fitted for the dynamic growth of the fruit, Is a randomly selected variable; Fitting curve for dynamic growth of fruits The acquisition mode of (a) is as follows: Randomly select Counting scaling points of the fitted curve by a counter when The points falling on the fruit dynamic growth fitting curve are less, the fitting effect is poor, and the selected fitting curve is discarded Continuing searching; Setting variable step length , The data pair formed by x and y is taken into the following formula: When n=1 of the number of the times, And so on, when the deviation of the observed value (x ', y') from the fitted curve is large, Circulating the belt into the above formula until When the description data pair basically falls on the fruit dynamic growth fitting curve; Obtaining a growth curve of Logistic, logarithmic, quadratic, cubic, linear fitting models respectively through a fruit dynamic growth fitting curve obtaining method, and analyzing and determining a curve model with highest fitness with an actual observation value; After obtaining the curve model with the highest fitness, the coloring period is used As dependent variable, the fitting value is obtained by using regression equation And using an accuracy factor And deviation factor The closer the deviation factor is to 1, the smaller the fluctuation amplitude of the fitting value and the observed value is, thereby proving the reliability of the curve model; 。
2. 2. the cherry fruit dynamic growth inversion method according to claim 1 is characterized by obtaining cherry fruit images in coloring period, specifically by mounting a camera on a support frame, keeping a fixed height, and controlling the shooting distance at 80 100Cm, wherein the shooting angle is controlled to be-15-90 degrees in the vertical direction and controlled to be in the horizontal direction And in the range of rotation, the focal length of the camera lens is adjusted, and the cherry fruit image in the coloring period is obtained.
3. 3. The cherry fruit dynamic growth inversion method according to claim 1, wherein the cherry fruit image is preprocessed to conform to the original shape of cherry fruit, specifically using 3 And 3, performing closed operation processing on the check cherry fruit image, filling the edge concave or middle cavity, so that the divided cherry fruit image is more in line with the original shape of the cherry fruit.
4. 4. The cherry fruit dynamic growth inversion method of claim 1, wherein the parameters are selected The fitting degree of the fruit dynamic growth fitting curve to the observed value is measured The larger the fitting effect is, the better, The closer the value of (2) is to 1, the better the fitting degree of the fitting curve to the observed value is; Wherein y is the observed value, Is the mean value, Is a fitting value.

Description

Cherry fruit dynamic growth inversion method Technical Field The invention relates to the technical field of greenhouse cherry cultivation, in particular to a cherry fruit dynamic growth inversion method. Background The method is a large agricultural country, the intelligent agricultural production, the digital agricultural management and the accurate agricultural service are necessarily developed. In recent years, the fruit industry in China develops particularly rapidly, the planting area and the yield expand rapidly, and the fruit industry has formed scale advantages and is continuously growing. With the establishment of agricultural information systems and the increase of large amounts of agricultural data, it is increasingly difficult to manually acquire and analyze data. Therefore, the data analysis technology is effectively applied in agriculture, provides practical information with predictability, time and guidance for farmers at proper time, realizes the transition from the traditional extensive planting management mode to the modern intelligent and scientific modes, and is beneficial to the rapid development of the traditional agriculture to the intelligent agriculture. Fruit growth is a complex process whose growth is nonlinear and mutable over time under the influence of various environmental conditions and random factors. However, cherry fruits are not drought-resistant and waterlogging-resistant, are sensitive to moisture, and are critical to watering in a proper period. The cherry fruit development process can be divided into a first rapid long period, a hard pit period and a second rapid long period, wherein the second rapid long period has the most vigorous cherry growth and development, is most sensitive to water supply, and determines the yield and quality of the fruits. During this period, the ovary cells divide vigorously, the cells expand rapidly, and the fruit size at harvest is mainly determined by the fruit development degree in a rapid and long period. In recent years, the popularization pace of large cherries is gradually accelerated, but systematic fruit growth and development dynamic research is lacking, and the change of environmental climate and imbalance of water and fertilizer management not only affect the growth and development of the fruits, but also affect the quality of the fruits, so that fruit dropping, fruit cracking, malformed fruits and the like are caused. Knowledge of cherry growth dynamics from fruiting to maturation is critical to farmer scientific water and fertilizer supply and water growth. However, monitoring cherry growth requires sizing the cherry fruit at each stage, which is a time consuming manual task. Disclosure of Invention The invention aims to provide a cherry fruit dynamic growth inversion method, which can be used for rapidly and accurately detecting the growth state of a fruit in a natural scene, dynamically grasping the growth rule of greenhouse crops and further improving the yield and quality of cherry fruits. In order to achieve the above purpose, the application provides a cherry fruit dynamic growth inversion method, which comprises the following steps: acquiring a cherry fruit image in a coloring period; preprocessing the cherry fruit image to enable the cherry fruit image to be more in line with the original shape of the cherry fruit; obtaining the projection area of cherry fruits by improving an intelligent scissors algorithm; Fitting a fruit dynamic growth curve according to an inversion method. Further, acquiring a coloring cherry fruit image, namely, mounting a camera on a support frame, keeping a fixed height, controlling a shooting distance to be 80-100 cm, controlling a shooting angle to be within a range of-15 DEG to 90 DEG in a vertical direction and 360 DEG in a horizontal direction, and adjusting the focal length of a camera lens to acquire the coloring cherry fruit image. Further, the cherry fruit image is preprocessed to be more in line with the original shape of the cherry fruit, specifically, 3 multiplied by 3 check cherry fruit images are adopted to perform closed operation processing, and edge concave or middle holes are filled, so that the segmented cherry fruit images are more in line with the original shape of the cherry fruit. Further, the projection area of cherry fruits is obtained by improving an intelligent scissors algorithm, and the method specifically comprises the following steps: firstly, acquiring a cherry fruit image under a natural environment obtained by opening and closing operation, detecting the edge of the image by using a Canny algorithm, finding out candidate circle centers, determining the radius according to the supporting degree of the edge of all the candidate circle centers on the cherry fruit image, and obtaining an optimal radius r; And performing semantic segmentation on the key feature areas of the cherry fruits pixel by combining a computer vision technology by taking an initial edge poi