CN100536653C - Crop water-requesting information determination based on computer vision - Google Patents

Abstract

A new method for detecting crop water demand information based on computer vision for intelligent water-saving irrigation. The crop water demand information detection system is composed of a black reference object of known size with proper size, an image acquisition device, an image acquisition card and a computer. It can realize non-contact, high-precision, and quick measurement of the size of crop stems or fruits, so as to obtain crop water demand information. Place a reference object of known size near the crop stem or fruit to be tested, take an image through an image acquisition device, perform a median filter on the image to remove noise, threshold the reference object and the object to be tested, and then calculate them separately The corresponding number of pixels, the ratio of the number of pixels between the two is equal to their real size ratio, so that the size of the crop stem or fruit can be obtained, and the water demand information of the crop can be obtained, and the irrigation system can be controlled to achieve the purpose of water-saving irrigation. The water saving rate is 30%, which can bring obvious economic and social benefits.

Description

A new method for detecting crop water demand information based on computer vision

Technical field

The invention relates to the field of water-saving irrigation and sensor measurement and control technology, in particular to a detection method for intelligent water-saving irrigation based on computer vision technology to detect micro-size changes of crop stems and fruits.

Background technique

The current irrigation system uses parameters such as soil humidity and environmental temperature as the control parameters of the irrigation system, not direct parameters, so the control accuracy is low, and the problem is that it causes a lot of waste of irrigation water resources.

In view of the above problems, a non-destructive plant water precision detection technology based on physical methods is proposed to realize the closed-loop control irrigation system, which is of great significance to improve the utilization efficiency of water resources in our country. In the early 1990s, researchers in our country used LVDT-5 (differential transformer type) displacement sensors to measure corn, citrus, etc., and achieved water saving effects of 15.6% and 21.4% respectively. However, LVDT is used in industrial metrology in the field of machinery manufacturing. The measurement resolution and accuracy can meet the requirements, but the indicators such as measurement force, volume and anti-environmental interference cannot meet the actual requirements for plant measurement. my country's orchard workers measure the size of citrus with an outer micrometer. During the growth period, when they find that the size of the fruit does not increase, they will irrigate the fruit tree in time. This method is time-consuming and laborious. Recently, scientists from Israel's Hebrew University also tested the tomato irrigation system using micron-level leaf thickness sensors, and achieved a water saving rate of 35% and an increase in yield of 40%, but it has not yet reached the level of practical application. At present, foreign scholars are also engaged in the use of short-range infrared remote sensing images to detect crop water requirements. This technology is very expensive and has not yet reached the level of practical application. "Water-saving irrigation system", by measuring the slight changes of plant organs through mechanical devices, and obtaining water demand information to control the irrigation system, the purpose of water saving can be achieved. This method of mechanical measurement requires contact measurement with mechanical probes, and every time The measurement must place the probe accurately on the original measurement position, or place the probe on the fruit or leaf in real time. The former is difficult to operate, and the latter will affect the growth of the crop due to the long-term placement of the mechanical probe, and the measurement pressure is difficult to control. If the pressure is too high, the crop organs will be crushed, which will affect the measurement accuracy; if the pressure is too small, the contact will be poor, which will make the measurement inaccurate.

At present, computer vision technology is developing rapidly, and the measurement accuracy is getting higher and higher. We propose to use this technology to measure crop stem changes or fruit growth in a non-contact manner, so as to detect crop water demand information and control the irrigation system to achieve water-saving irrigation. This technology is a new type of sensor that meets the needs of plant detection. This research can serve my country's water-saving irrigation as early as possible, and has excellent innovation and practicability.

Contents of the invention

The purpose of the present invention is to provide an intelligent water-saving irrigation detection method based on computer vision technology to detect the micro-size changes of crop stems and fruits, so as to realize non-contact measurement without affecting the growth of crops and the purpose of real water-saving irrigation.

In order to achieve the above object, the technical scheme adopted in the present invention is as follows:

Scheme 1: It consists of image acquisition equipment, image acquisition card and computer; image acquisition equipment is used to collect images of crop stems. In order to solve the problem of complicated calibration in measurement, we propose the method of using reference objects, which is easy to operate and accurate in measurement High, that is to choose a reference object with an appropriate size and a known size, and place the reference object in the same plane with the same distance from the stem of the crop to be measured to the lens, so that the collected image contains both the stem of the crop to be measured and the crop that has been measured. The reference object of known size, through the image processing algorithm, first segment the crop stem and the reference object, and then calculate the number of pixels occupied in the direction of their diameter respectively, the ratio of the two is equal to the size ratio of their real diameter, so that The stalk of the crop can be measured, so as to obtain the water demand of the crop and achieve the purpose of water-saving irrigation.

Scheme 2: It is composed of image acquisition equipment, image acquisition card and computer; image acquisition equipment is used to collect crop fruit images. In order to solve the problem of complicated calibration in measurement, we propose the method of using reference objects, which is easy to operate and has high measurement accuracy , that is to choose a reference object with an appropriate size and a known size, and place the reference object in the same plane with the same distance from the crop fruit to be measured to the lens, so that the collected image contains both the crop fruit to be measured and the fruit of known size The reference object, through the image processing algorithm, first divides the crop fruit and the reference object, and then calculates the number of pixels occupied by their diameters respectively. The ratio of the two is equal to the size ratio of their real diameters, so that the crops can be measured The diameter of the fruit, so as to obtain the water demand of the crops, to achieve the purpose of water-saving irrigation.

Scheme 3: It is composed of image acquisition equipment, image acquisition card and computer; image acquisition equipment is used to collect crop fruit images. In order to solve the problem of complicated calibration in measurement, we propose the method of using reference objects, which is easy to operate and has high measurement accuracy , that is to choose a reference object with an appropriate size and a known size, and place the reference object in the same plane with the same distance from the crop fruit to be measured to the lens, so that the collected image contains both the crop fruit to be measured and the fruit of known size The reference object, through the image processing algorithm, first segment the crop fruit and the reference object, and then calculate the number of pixels contained in their area respectively, the ratio of the two is equal to their area ratio, so that the fruit size of the crop can be measured, In order to obtain the water demand of the crops, the purpose of water-saving irrigation can be achieved.

Scheme 4: It is composed of image acquisition equipment, image acquisition card and computer; the image acquisition equipment is used to collect images of crop stems or fruits. High measurement accuracy, that is, choose a reference object with an appropriate size and a known size. If the size of the selected reference object is larger than the object to be measured, place the reference object behind the object to be measured. If the size of the selected reference object is smaller than the object to be measured, The reference object is placed in front of the object to be measured, so that the collected image contains both the fruit of the crop to be measured and a reference object of known size. Through the above three image processing algorithms, the stem of the crop can be measured. The size of the stalk or fruit can be used to obtain the water demand of the crops and achieve the purpose of water-saving irrigation.

Scheme 5: It is composed of image acquisition equipment, image acquisition card and computer; when the stem or fruit size of the crop is too large, according to the measurement principles of the first four schemes, the resolution of the image acquisition equipment is required to be relatively high, so that the measurement equipment The cost will increase rapidly or the measurement cannot be realized. For this reason, we adopt the following method: use image acquisition equipment to collect images of crop stems or fruits. In order to solve the problem of complicated calibration in measurement, we propose the method of using reference objects, which is simple And the measurement accuracy is high, that is, choose a reference object with a known size of appropriate size. If the size of the selected reference object is larger than the object to be measured, place the reference object behind the object to be measured. If the size of the selected reference object is smaller than the object to be measured place the reference object in front of the object to be measured, and then under the same conditions, collect the images of the reference object and the left and right edges of the object to be measured, and then use the image processing algorithm to obtain the reference object in the two images respectively. The number of pixels that differ from the edge of the object to be measured, so that the difference in diameter between the two is proportional to the sum of the number of pixels that differ at both ends, and the diameter of the stem or fruit of the crop to be measured can be obtained, thereby obtaining the water requirement of the crop. situation, to achieve the purpose of water-saving irrigation.

Compared with the background technology, the present invention has the beneficial effects of:

(1) Realize non-contact measurement, simple operation, high measurement accuracy, does not affect the normal growth of crops, and achieves the same expected water saving rate as Chinese patent 03150690.9;

(2) Applicable to a wide range of crops, it can be installed in indoor and outdoor irrigation systems such as smart greenhouses and modern agricultural bases, and is suitable for various vegetables, corn, wheat, soybeans, and fruit saplings;

Obviously, compared with changes in soil moisture, the morphological or physiological changes of plant organs such as leaves, stems, and fruits can more directly, comprehensively, quickly, and sensitively reflect the water conditions in plants. In this way, the diagnosis of plant water needs can be transformed into the research of the measurement accuracy of the micro-displacement sensor and the corresponding laws. The irrigation system composed in this way is an optimal control system, which can improve the utilization efficiency of irrigation water, that is, achieve the purpose of saving water.

Description of drawings

Fig. 1 is a block diagram of the composition of the measurement system of the present invention;

Figure 2 crop stem image;

Fig. 3 is a block diagram of a processing method for measuring crop stems or fruits.

In Figure 2, (a) is the collected image of crop stems and reference objects; (b) is the binary image of crop stems and reference objects after filtering, segmentation, removing noise points and filling holes

Detailed ways

In order to better understand the technical solution of the present invention, the implementation of the present invention will be further described below in conjunction with the embodiment of measuring the stems of crops.

According to the specific measurement object is the crop stalk, as shown in Figure 1, the measurement device is constructed. Choose the appropriate image acquisition device (PUNIX-7DSP camera) and the corresponding image acquisition card (MatroxII-MC4) according to the size of the crop stem. According to the collected image segmentation experiment, we choose a black reference object that is easier to achieve segmentation, and the collected image is shown in Figure 2(a).

The block diagram of the processing method is shown in Figure 3. First, the image is filtered to remove noise to improve the measurement accuracy, and then the image histogram is obtained, and the segmentation threshold is automatically determined according to the histogram to segment the image to obtain reference objects and crop stems, and then impurities are removed. And fill the hole, and finally calculate the pixel ratio of the crop stem and the reference object to get the change of the crop stem, so as to get whether the crop needs water, guide the irrigation system, and achieve the purpose of water saving. The specific implementation details of each part are as follows:

1. Use median filter to remove noise

The basic idea of median filtering is to replace the gray value of the pixel with the median value of the gray value of the neighborhood of the pixel. This method can preserve the edge details of the image while removing impulse noise and salt and pepper noise. We filtered the acquired images with a 3×3 window.

2. Automatic determination of threshold

We use the method of finding the extreme value to determine the threshold value of the segmented image, that is, first obtain the histogram function f(n) of the image, n is the gray level 1 to 255, and calculate the first-order derivative function f'( n) and second-order derivative function f "(n). When the first-order derivative f'(k)=0 of a certain point k therein, it means that the k point is a stagnation point, and at this moment,

If f″(k)>0, point k is the minimum value, which is the valley between the two peaks of the histogram; if f″(k)<0, then point k is the maximum value, which is the bottom of the histogram crest.

Since the data of the numerical image is discrete, the first derivative function is:

$\begin{array}{cc}\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; \&prime;</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; h</span>}}& \mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; ></span>\mathrm{<span\; class="notranslate">\; 0</span>}\end{array}$

The second derivative function is:

$\begin{array}{cc}\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; \&prime;</span><span\; class="notranslate">\; \&prime;</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; \&prime;</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; \&prime;</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; h</span>}}& \mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; ></span>\mathrm{<span\; class="notranslate">\; 0</span>}\end{array}$

According to the experiment, this research uses the first minimum value and the second minimum value to segment the reference object, and then uses the second and third minimum values to segment the crop stems.

3. Removal of impurities and filling of voids

After the reference object and crop stems are segmented, the object contains holes, and the image contains many noise points, which will affect the subsequent measurement. We judge whether there are more than five of the eight adjacent pixels of each pixel. The pixel has a different value, if so, adjust the grayscale of the pixel to the color value of the majority of pixels.

4. Calculate the reference object and the number of pixels in the diameter direction of the stem to obtain the stem

In order to further improve the measurement accuracy, we use the least squares method to fit the edges of the reference object and the crop stem, and then calculate their diameter ratio to obtain the size of the crop stem and thus the water demand of the crop.

Take an edge as an example, set the number of edge pixels as N,

Calculate the coordinates of the geometric center of this edge: $Y_a=\frac{1}{N}\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}{Y}_i,$ $x_a=\frac{1}{N}\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}{x}_i---\left(1\right)$

Calculate the slope: $B_a=[\underset{i=1}{\overset{N/2}{\mathrm{\&Sigma;}}}{Y}_i-\underset{i=N/2-1}{\overset{N}{\mathrm{\&Sigma;}}}{Y}_i]/\left[[\underset{i=1}{\overset{N/2}{\mathrm{\&Sigma;}}}{x}_i-\underset{i=N/2-1}{\overset{N}{\mathrm{\&Sigma;}}}{x}_i]\right]---\left(2\right)$

Calculate the intercept: C _a =Y _a -B _a ×X _a (3)

The equation of the edge fitting line is: Y=B _a ×X+C _a (4)

Then, another edge fitting equation is: Y=B _b ×X+C _b (5)

Since the two edges are approximately parallel, the slope is B=(B _a +B _b )/2 the common slope of the two edges. In this way, a vertical line is made through a certain point of one of the edges to obtain the point intersecting with the other edge, and the distance between the two points is calculated to be the number of pixels corresponding to the diameter of the reference object or the crop stem. The ratio of the diameter of the reference object to the crop stem can be obtained. According to the known diameter of the reference object, the diameter of the stem can be obtained. The change of the crop stem in a day reflects the water demand information of the crop, thereby guiding irrigation and saving water.

Claims (1)

1. based on the crop water-requesting information determination of computer vision, detection system is made up of the suitable reference substance of size, image capture device, image pick-up card and calculator; It is characterized in that object of reference is placed near the crop cane and fruit to be measured, obtain crop cane size and fruit size, obtain crop and need water information by corresponding image processing algorithm, thus the control irrigation system, and described image processing algorithm comprises:

(1) medium filtering is removed noise: adopt the medium filtering of 3 * 3 windows that the image of gathering is carried out filtering removal noise, improve certainty of measurement;

(2) determining automatically of threshold value: the threshold value of determining split image with the method for asking extreme value, promptly try to achieve the histogram functions of image earlier, calculate its single order derived function and second order derived function again, judge all minimums, with first minimum and second minimum object of reference is split, with second and the 3rd minimum crop cane or fruit are split again;

(3) filling in the removal of impure point and cavity: after object of reference and crop cane or fruit split, by whether having in adjacent eight pixels of judging each pixel more than five and this pixel different value, if then this pixel grey scale is adjusted into the color value of most pixels;

(4) pixel count that calculates object of reference and cane or fruit obtains cane or fruit size: utilize method of least squares to come the edge of match object of reference and crop cane, try to achieve out their diameter ratio then, just can obtain the size of crop cane or fruit, thereby obtain crop and need aqueous condition, perhaps when crop cane and fruit are bigger, by measuring the size that pixel quantity that object of reference and determinand edges at two ends differ obtains crop cane or fruit, thereby obtain crop and need water information, control irrigation system, reach the purpose of water saving.

Images