Garden maintenance information processing system based on image processing

Through image processing technology, including image acquisition, image processing and neural network recognition, the feature blur problem caused by plants of similar colors in the garden is solved, and the accurate identification and management of the vegetation range of the garden is achieved.

CN118628752BActive Publication Date: 2024-11-22WUHAN TONGCHUANG WANZHI DIGITAL TECH CO LTD
View PDF 8 Cites 0 Cited by

Patent Information

Application Number
CN202411096369.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-12
Publication Date
2024-11-22
Estimated Expiration
2044-08-12

AI Technical Summary

Technical Problem

Plants with similar colors in the garden cause blurred features, and traditional methods are difficult to accurately identify the vegetation range from images.

Method used

Using an image processing-based system, the garden image is acquired through the image acquisition module. The image processing module integrates the three-dimensional spatial distribution comparison coordinate system to generate homomorphic layered images, calculate the color-oriented consistency of the connecting domain, and perform round-rotation growth to identify the same color Domain area and use neural network to identify garden area information.

Benefits of technology

Accurate identification of plants of similar colors in the garden is achieved, the accuracy and efficiency of regional growth is improved, and more accurate vegetation range detection is obtained.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118628752B_ABST
    Figure CN118628752B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of image processing technology, and in particular to a garden maintenance information processing system based on image processing, including: shooting an image to be processed; performing channel fusion processing on the image to be processed to obtain a number of homomorphic layered images; performing region growth on the homomorphic layered images to obtain each homochromatic region; mapping each homochromatic region to the image to be processed to obtain each detection region; and performing neural network recognition on the detection region to realize garden maintenance information management. The present invention uses the color change information generated by the gaps between plants and the color difference of the surrounding plants to help determine the specific edge of the object, reducing the problem of mismatching caused by the presence of more plants with similar colors in the garden, and can obtain plant information and perform information management more accurately, comprehensively, conveniently and intelligently.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of image processing, and in particular to a garden maintenance information processing system based on image processing. Background Art

[0002] The application of Internet of Things technology and artificial intelligence algorithms to develop intelligent maintenance equipment, such as automatic sprinkler irrigation systems and automatic pruning equipment, can realize automated and intelligent garden maintenance operations and improve work efficiency and accuracy. However, the similar textures of plants with similar colors in the garden will lead to blurred features, and the traditional regional growth method cannot identify the exact vegetation range from the image. Summary of the invention

[0003] In order to solve the above problems, the present invention provides a garden maintenance information processing system based on image processing.

[0004] The garden maintenance information processing system based on image processing of the present invention adopts the following technical solutions:

[0005] An embodiment of the present invention provides a garden maintenance information processing system based on image processing, the system comprising:

[0006] An image acquisition module is used to use unmanned patrol equipment to capture key areas of the garden to obtain images to be processed;

[0007] The image processing module is used to integrate different color spaces of the image to be processed to obtain a three-dimensional Spatial distribution comparison coordinate system, using three-dimensional A plurality of homomorphic layered images are obtained by comparing the spatial distribution with the coordinate system, and the growth seed points of each connected domain on the refined homomorphic image are obtained on the homomorphic layered images. The color-oriented consistency amount of each connected domain is calculated based on the growth seed points of each connected domain on the refined homomorphic image, and each homochromic domain area is obtained by performing round growth according to the color-oriented consistency amount of each connected domain, and each homochromic domain area is mapped to the image to be processed to obtain each detection area;

[0008] The information recognition module is used to obtain the regional information of the key focus areas of the garden through neural network recognition based on all the detection areas.

[0009] Preferably, the three-dimensional image is obtained by integrating different color spaces of the image to be processed. The spatial distribution reference coordinate system includes the following specific steps:

[0010] All pixels in the image to be processed are numbered first, and the The images of the three channels are extracted respectively to obtain three single-channel images, and each single-channel image has a pixel number corresponding to that in the image to be processed. On each single-channel image, the pixel position distribution is used as the bottom coordinate, and the pixel value of the pixel at each position is used as the height to construct a single-channel three-dimensional coordinate system; the three single-channel three-dimensional coordinate systems are aligned according to their respective pixel numbers: the bottom origin coordinates of the three single-channel three-dimensional coordinate systems are aligned, and the bottom coordinates of the three single-channel three-dimensional coordinate systems are respectively represented and Plane, merge the heights of the three single-channel 3D coordinate systems and The spatial area of ​​the plane envelope is used as the internal pixel points of the space to establish the three-dimensional image to be processed. Spatial distribution comparison coordinate system.

[0011] Preferably, the use of three-dimensional The spatial distribution is compared with the coordinate system to obtain a number of homogeneous layered images, including the following specific steps:

[0012] Three-dimensional All the internal pixels of the spatial distribution reference coordinate system are clustered by distance using a clustering algorithm to obtain several clusters. The corresponding positions of the pixels of each cluster on the image to be processed are projected onto a blank image, and the grayscale values ​​of the pixels at the projected positions on the blank image are set to 255, and the grayscale values ​​of the pixels at the remaining positions are set to 0, to obtain a homomorphic layered image of each cluster.

[0013] Preferably, the step of obtaining the growth seed points of each connected domain on the refined homomorphic layered image comprises the following specific steps:

[0014] On each homomorphic layered image, the opening operation in morphological processing is used to open each homomorphic layered image to obtain a refined homomorphic image. The number of connected domains of all pixels with a grayscale value of 255 in the refined homomorphic image is counted, and the centroid of each connected domain is recorded as the growth seed point of each connected domain.

[0015] Preferably, the color-guided consistency amount of each connected domain is calculated based on the growth seed point of each connected domain on the refined homomorphic image, and the specific steps include the following:

[0016] On each refined homomorphic image, the number of all edge pixels in each connected domain, the growth vector of each edge pixel in the edge pixel sequence, and the overall growth distance are obtained. The color-oriented consistency of each connected domain is calculated according to the number of all edge pixels in each connected domain, the growth vector of each edge pixel in the edge pixel sequence, and the overall growth distance. The specific calculation formula is as follows:

[0017]

[0018] in, Indicates The color-oriented consistency of the connected domain, Indicates The number of all edge pixels in a connected domain, and Respectively represent The first pixel sequence of the edge pixels of the connected component edge pixels and the The growth vector of edge pixels, and , Indicates calculating the cosine similarity of the two vectors in the brackets. Indicates The overall growth distance of the connected domain.

[0019] Preferably, the obtaining of the number of all edge pixels of each connected domain, the growth vector of each edge pixel in the edge pixel sequence, and the overall growth distance comprises the following specific steps:

[0020] The edge pixel points of each connected domain are obtained on each refined homomorphic image. For any connected domain, the number of all edge pixel points of the connected domain is counted, and all edge pixel points of the connected domain are sorted in a clockwise direction starting from any edge pixel point to obtain the edge pixel point sequence of the connected domain. The vectors of all edge pixel points in the edge pixel point sequence of the connected domain pointing to the growth seed point of the connected domain are counted, and each of the vectors is recorded as the growth vector of each edge pixel point, and the arithmetic mean of the module lengths of all growth vectors is recorded as the overall growth distance of the connected domain.

[0021] Preferably, the step of performing round growth according to the color-oriented consistency amount of each connected domain to obtain each same-color domain region includes the following specific steps:

[0022] Each connected domain is recorded as an initial connected domain, and each initial connected domain is grown in rounds. In each round of growth, all pixels outside all edge pixels of the connected domain and within the eight neighborhoods of the edge pixels are taken as new edge pixels, and all new edge pixels are added to each initial connected domain to form a new connected domain. The color-oriented consistency of the new connected domain is calculated, and it is determined whether the color-oriented consistency of the new connected domain satisfies both the mutation condition and the continuous orientation condition. If so, the next round of growth is performed to obtain an updated connected domain, and it is determined whether the color-oriented consistency of the updated connected domain satisfies both the mutation condition and the continuous orientation condition, and so on, until the color-oriented consistency of the updated connected domain does not satisfy both the mutation condition and the continuous orientation condition, and the connected domain of the last round of growth is recorded as each same-color domain region.

[0023] Preferably, the mutation condition specifically refers to:

[0024] A mutation threshold is preset. When the color-oriented consistency amount of the new connected domain is less than or equal to the mutation threshold, the mutation condition is met. When the color-oriented consistency amount of the new connected domain is greater than the mutation threshold, the mutation condition is not met.

[0025] Preferably, the continuous guiding condition specifically refers to:

[0026] The preceding continuity factor of the new connected domain is obtained, and the continuity guidance coefficient is preset. When the preceding continuity factor of the new connected domain is less than or equal to the continuity guidance coefficient, the continuity guidance condition is satisfied. When the preceding continuity factor of the new connected domain is greater than the continuity guidance coefficient, the continuity guidance condition is not satisfied.

[0027] Preferably, the obtaining of the preceding continuity factor of the new connected domain comprises the following specific steps:

[0028] Starting from the initial connected domain, the color-oriented consistency amount of the connected domain is stored in a set, and the set is recorded as a loop parameter set. Each time the color-oriented consistency amount of a new connected domain is calculated, the color-oriented consistency amount of the new connected domain is put into the loop parameter set and the arithmetic mean of all the color-oriented consistency amounts in the loop parameter set is calculated, and the arithmetic mean is recorded as the leading continuity factor of the new connected domain.

[0029] The beneficial effect of the technical solution of the present invention is: in view of the technical problem that similar textures of plants with similar colors in gardens will lead to blurred features, and the traditional region growing method cannot distinguish the accurate vegetation range from the image, the present invention analyzes the edge guide features generated when the gradient edge is processed by the integrated region growing algorithm through analyzing the plant growth characteristics, and performs a comprehensive analysis of color information and position information. When the edge generates the guide region guidance, the region growth is stopped, a more accurate detection area is generated, and a more accurate regional situation is obtained. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0031] Figure 1 The present invention is a structural composition diagram of the garden maintenance information processing system based on image processing. DETAILED DESCRIPTION

[0032] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation method, structure, features and effects of the garden maintenance information processing system based on image processing proposed by the present invention are described in detail below in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0033] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0034] The specific scheme of the garden maintenance information processing system based on image processing provided by the present invention is described in detail below with reference to the accompanying drawings.

[0035] See also Figure 1 , which shows a structural composition diagram of a garden maintenance information processing system based on image processing provided by an embodiment of the present invention, the system comprising:

[0036] The image acquisition module is used to use unmanned patrol equipment to capture key areas of the garden to obtain images to be processed.

[0037] It should be noted that for the garden maintenance information management system, garden management personnel will pre-set the more important areas in the garden as key focus areas. The growth of plants in the key focus areas is generally monitored by real-time aerial photography by drones. Therefore, it is necessary to capture real-time images and perform feature extraction through electronic equipment.

[0038] Specifically, at any time, the key areas of the garden are photographed by electronic equipment to obtain a number of color images, and each color image is used as an image to be processed.

[0039] The electronic device includes: a camera, a drone equipped with the camera, and a drone control device.

[0040] At this point, the image to be processed is obtained.

[0041] The image processing module is used to integrate different color spaces of the image to be processed to obtain a three-dimensional Spatial distribution comparison coordinate system, using three-dimensional A number of homomorphic layered images are obtained by comparing the spatial distribution with the coordinate system. The growth seed points of each connected domain on the refined homomorphic image are obtained on the homomorphic layered images. The color-guided consistency amount of each connected domain is calculated based on the growth seed points of each connected domain on the refined homomorphic image. According to the color-guided consistency amount of each connected domain, rounds of growth are performed to obtain each homochromatic region. Each homochromatic region is mapped to the image to be processed to obtain each detection region.

[0042] It should be noted that since the growth conditions of plants in the key focus areas of the garden involve the acquisition of information on multiple colors, and according to common sense, plants of the same type have more similar position distribution and color distribution in the image to be processed, it is necessary to combine the information of the image in different color channels, and combine this information with the position information to obtain information that can simultaneously reflect both color and position information, that is, the homomorphic layered image used in this embodiment, and based on the information on each homomorphic layered image, classify different types of plants in the original image, and divide different detection areas that may represent information on different types of plants for subsequent operations.

[0043] Specifically, all the pixels in the image to be processed are numbered first, and the pixels of the image to be processed are The images of the three channels are extracted respectively to obtain three single-channel images, and each single-channel image has a pixel number corresponding to that in the image to be processed. On each single-channel image, the pixel position distribution is used as the bottom coordinate, and the pixel value of the pixel at each position is used as the height to construct a single-channel three-dimensional coordinate system; the three single-channel three-dimensional coordinate systems are aligned according to their respective pixel numbers: the bottom origin coordinates of the three single-channel three-dimensional coordinate systems are aligned, and the bottom coordinates of the three single-channel three-dimensional coordinate systems are respectively represented and Plane, merge the heights of the three single-channel 3D coordinate systems and The spatial area of ​​the plane envelope is used as the internal pixel points of the space to establish the three-dimensional image to be processed. Spatial distribution comparison coordinate system.

[0044] Furthermore, for three-dimensional All the internal pixels of the spatial distribution comparison coordinate system are used The algorithm performs distance clustering to obtain several clusters. The corresponding positions of the pixels of each cluster on the image to be processed are projected onto a blank image. The grayscale values ​​of the pixels at the projected positions on the blank image are set to 255, and the grayscale values ​​of the pixels at the remaining positions are set to 0. This obtains a homomorphic layered image for each cluster, and further obtains several homomorphic layered images.

[0045] It should be further explained that the above three-dimensional The spatial distribution is compared with the coordinate system to obtain a homogeneous layered image, which is the information that reflects both color and position information as mentioned above. The homogeneous layered image actually reflects the specific distribution position of each type of pixel points with similar colors in space, which is equivalent to mapping the color information to the position information. In the homogeneous layered image, regional growth can be performed on pixels with similar color information to obtain the best unified measurement standard for their position and color information, that is, the same color domain area.

[0046] Specifically, on each homomorphic layered image, the erosion algorithm in morphological processing is used to eliminate the edge adhesion areas in the image to obtain a refined homomorphic image, and the number of connected domains of all pixels with a grayscale value of 255 in the refined homomorphic image is counted, and the centroid of each connected domain is recorded as the growth seed point of each connected domain.

[0047] It should be further explained that when performing region growing on each connected domain, it is necessary to consider the overall difference of the edge pixels of each connected domain on the refined homomorphic image. During the growing process, if the grown region cannot guarantee the unified measurement of position and color information, that is, the overall difference of the edge pixels of each connected domain is too large, it is considered that it does not actually represent a type of plant, and this should be used as a basis during the growth process.

[0048] Specifically, the edge pixel points of each connected domain on each refined homomorphic image are obtained. For any connected domain, the number of all edge pixel points of any connected domain is counted, and all edge pixel points of any connected domain are sorted in a clockwise direction starting from any initial pixel point to obtain the edge pixel point sequence of any connected domain. The vectors of all edge pixel points pointing to the growth seed point of the connected domain in the edge pixel point sequence of any connected domain are counted, and each of the vectors is recorded as the growth vector of each edge pixel point, and the arithmetic mean of the modulus length of all vectors is taken as the overall growth distance of each connected domain. Feature extraction is performed on each connected domain to obtain the color-guided consistency amount of the connected domain. The specific calculation formula is as follows:

[0049]

[0050] in, Indicates The color-oriented consistency of the connected domain, Indicates The number of all edge pixels in a connected domain, and Respectively represent The first pixel sequence of the edge pixels of the connected component edge pixels and the The growth vector of edge pixels, and , Indicates calculating the cosine similarity of the two vectors in the brackets. Indicates The overall growth distance of a connected domain. When the overall growth distance of a connected domain is large, it means that the overall position information of the connected domain is quite different during the regional growth process. This factor should be weakened as much as possible during the growth process. Similarly, when the sum of the growth vectors of all adjacent edge pixels in the connected domain is large, it is considered that the connected domain has a tendency in a certain direction, indicating that the position distribution of a type of plant represented by it in the actual space is irregular. This factor should also be weakened as much as possible during the growth process.

[0051] It should be further explained that when performing traditional regional growing, only mutation conditions need to be set. However, within the region of regional growth, the pixel point is reflected as the center of the vegetation connected domain. The pixels have more similar pixel components in the color space, but the fluctuation direction is not consistent. Therefore, when the current growth cycle is small, the color distance vector directions of adjacent pixels at the edge of the region are more different, indicating that the grayscale value of the plant region itself is relatively uniform. Even if there is a small fluctuation in pixel distance, the overall situation is still highly similar. Therefore, it is necessary to consider continuous guidance conditions to solve this problem to improve the rationality of regional growth.

[0052] Specifically, each connected domain is recorded as an initial connected domain, and each initial connected domain is grown in rounds. In each round of growth, all pixels outside all edge pixels of the connected domain and within the eight neighborhoods of the edge pixels are taken as new edge pixels, and all new edge pixels are added to each initial connected domain to form a new connected domain. Feature extraction is performed on the new connected domain to obtain the color-oriented consistency amount of the new connected domain, and it is determined whether the color-oriented consistency amount of the new connected domain satisfies both the mutation condition and the continuous orientation condition. If so, the next round of growth is performed to obtain an updated connected domain, and it is determined whether the color-oriented consistency amount of the updated connected domain satisfies both the mutation condition and the continuous orientation condition, and so on, until the color-oriented consistency amount of the updated connected domain does not satisfy both the mutation condition and the continuous orientation condition, and the connected domain of the last round of growth is recorded as each same-color domain area.

[0053] The mutation condition is: a mutation threshold is preset, when the color orientation consistency of the new connected domain is less than or equal to the mutation threshold, the mutation condition is met, and when the color orientation consistency of the new connected domain is greater than the mutation threshold, the mutation condition is not met. This embodiment takes the mutation threshold of 5 as an example for explanation, and does not specifically limit it.

[0054] The continuity guidance condition is as follows: starting from the initial connected domain, the color guidance consistency of the connected domain is stored in a set, and the set is recorded as a loop parameter set. Each time the color guidance consistency of the new or updated connected domain is calculated, the color guidance consistency of the new or updated connected domain is placed in the loop parameter set and the arithmetic mean of all the color guidance consistency in the loop parameter set is calculated, and the arithmetic mean is recorded as the pre-continuity factor of the new or updated connected domain. When the pre-continuity factor of the new connected domain is less than or equal to the continuity guidance coefficient, the continuity guidance condition is satisfied. When the pre-continuity factor of the updated connected domain is greater than the continuity guidance coefficient, the continuity guidance condition is not satisfied. This embodiment is described by taking the continuity guidance coefficient of 1.5 as an example, and no specific limitation is imposed on it.

[0055] It should be further explained that each isochromatic region exists in each refined isomorphic image and is the independent distribution characteristic of various plants. If the image to be processed is to be integrated as a whole, it is necessary to feed it back to the image to be processed according to the position information of the original channel fusion method in order to extract the final overall information.

[0056] Specifically, all pixels in each homochromatic region on each refined homomorphic image are classified into one category, and the pixel values ​​of each category are found in the three-dimensional The spatial distribution is compared with the corresponding position in the coordinate system, and further based on the three-dimensional The spatial distribution is compared with the corresponding position in the coordinate system to obtain the position of each type of pixel on the image to be processed, and all the classes on the image to be processed are counted, and each class is recorded as each detection area.

[0057] So far, several detection areas have been obtained.

[0058] The information recognition module is used to obtain the regional information of the key focus areas of the garden through neural network recognition based on all the detection areas.

[0059] It should be noted that after the detection area is acquired, the information including the plant species information contained in the detection area can be extracted and represented as data text using a neural network based on a long-term and large-scale data set as a benchmark, and then relevant personnel can make subsequent management arrangements.

[0060] Specifically, all detection areas are input into the trained neural network to obtain the regional information of the key focus areas of the garden, the regional information of the key focus areas of the garden is summarized to the garden control center and the relevant personnel conduct regular analysis of the information, and the garden is planned and designed according to the analysis results to realize the information management of garden maintenance.

[0061] In this embodiment, the neural network adopts an Encoder-Decoder structure;

[0062] The training method of the trained neural network used in this embodiment is as follows:

[0063] Collect a batch of images of key areas of interest in the garden in advance, and provide the corresponding annotations of the plant species pre-planted in the garden for each image. Ensure that the images and annotations are paired; obtain a number of detection areas from a batch of images of key areas of interest in the garden according to the steps described in this embodiment, use these detection areas and corresponding annotations as a data set, use the data set to train a semantic segmentation network as a trained neural network, and record the obtained data as the regional information of the key areas of interest in the garden. It should be noted that the training method of the neural network is a well-known technology, and will not be described in detail in this embodiment.

[0064] At this point, the information management of garden maintenance has been realized.

[0065] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. The garden maintenance information processing system based on image processing is characterized by: The system includes: An image acquisition module is used to use unmanned patrol equipment to capture key areas of the garden to obtain images to be processed; The image processing module is used to integrate different color spaces of the image to be processed to obtain a three-dimensional Spatial distribution comparison coordinate system, using three-dimensional A plurality of homomorphic layered images are obtained by comparing the spatial distribution with the coordinate system, and the growth seed points of each connected domain on the refined homomorphic image are obtained on the homomorphic layered images. The color-oriented consistency amount of each connected domain is calculated based on the growth seed points of each connected domain on the refined homomorphic image, and each homochromic domain area is obtained by performing round growth according to the color-oriented consistency amount of each connected domain, and each homochromic domain area is mapped to the image to be processed to obtain each detection area; An information recognition module is used to obtain regional information of key areas of the garden through neural network recognition based on all detection areas; The use of three-dimensional The spatial distribution is compared with the coordinate system to obtain a number of homogeneous layered images, including the following specific steps: Three-dimensional All the internal pixels of the spatial distribution reference coordinate system are clustered by distance using a clustering algorithm to obtain several clusters. The corresponding positions of the pixels of each cluster on the image to be processed are projected onto a blank image. The grayscale values ​​of the pixels at the projected positions on the blank image are set to 255, and the grayscale values ​​of the pixels at the remaining positions are set to 0, to obtain a homomorphic layered image of each cluster. The specific steps of performing round growth according to the color-oriented consistency amount of each connected domain to obtain each same-color domain region are as follows: Each connected domain is recorded as an initial connected domain, and each initial connected domain is grown in rounds. In each round of growth, all pixels outside all edge pixels of the connected domain and within the eight neighborhoods of the edge pixels are taken as new edge pixels, and all new edge pixels are added to each initial connected domain to form a new connected domain. The color-oriented consistency of the new connected domain is calculated, and it is determined whether the color-oriented consistency of the new connected domain satisfies both the mutation condition and the continuous orientation condition. If so, the next round of growth is performed to obtain an updated connected domain, and it is determined whether the color-oriented consistency of the updated connected domain satisfies both the mutation condition and the continuous orientation condition, and so on, until the color-oriented consistency of the updated connected domain does not satisfy both the mutation condition and the continuous orientation condition, and the connected domain of the last round of growth is recorded as each same-color domain region.

2. The garden maintenance information processing system based on image processing according to claim 1 is characterized in that: The method integrates different color spaces of the image to be processed to obtain a three-dimensional The spatial distribution reference coordinate system includes the following specific steps: All pixels in the image to be processed are numbered first, and the The images of the three channels are extracted respectively to obtain three single-channel images, and each single-channel image has a pixel number corresponding to the pixel number in the image to be processed. On each single-channel image, the pixel position distribution is used as the bottom coordinate and the pixel value of the pixel at each position is used as the height to construct a single-channel three-dimensional coordinate system; Align the three single-channel 3D coordinate systems according to their respective pixel numbers: Align the bottom origin coordinates of the three single-channel 3D coordinate systems, and use the bottom coordinates of the three single-channel 3D coordinate systems to represent and Plane, merge the heights of the three single-channel 3D coordinate systems and The spatial area of ​​the plane envelope is used as the internal pixel points of the space to establish the three-dimensional image to be processed. Spatial distribution comparison coordinate system.

3. The garden maintenance information processing system based on image processing according to claim 1 is characterized in that: The specific steps of obtaining the growth seed points of each connected domain on the refined homomorphic image on the homomorphic layered image are as follows: On each homomorphic layered image, the opening operation in morphological processing is used to open each homomorphic layered image to obtain a refined homomorphic image. The number of connected domains of all pixels with a grayscale value of 255 in the refined homomorphic image is counted, and the centroid of each connected domain is recorded as the growth seed point of each connected domain.

4. The garden maintenance information processing system based on image processing according to claim 1 is characterized in that: The specific steps of calculating the color-guided consistency amount of each connected domain based on the growth seed point of each connected domain on the refined homomorphic image are as follows: On each refined homomorphic image, the number of all edge pixels in each connected domain, the growth vector of each edge pixel in the edge pixel sequence, and the overall growth distance are obtained. The color-oriented consistency of each connected domain is calculated according to the number of all edge pixels in each connected domain, the growth vector of each edge pixel in the edge pixel sequence, and the overall growth distance. The specific calculation formula is as follows: in, Indicates The color-oriented consistency of the connected domain, Indicates The number of all edge pixels in a connected domain, and Respectively represent The first pixel sequence of the edge pixels of the connected component edge pixels and the The growth vector of edge pixels, and , Indicates calculating the cosine similarity of the two vectors in the brackets. Indicates The overall growth distance of the connected domain.

5. The garden maintenance information processing system based on image processing according to claim 4 is characterized in that: The specific steps of obtaining the number of all edge pixels of each connected domain, the growth vector of each edge pixel in the edge pixel sequence, and the overall growth distance are as follows: The edge pixel points of each connected domain are obtained on each refined homomorphic image. For any connected domain, the number of all edge pixel points of the connected domain is counted, and all edge pixel points of the connected domain are sorted in a clockwise direction starting from any edge pixel point to obtain the edge pixel point sequence of the connected domain. The vectors of all edge pixel points in the edge pixel point sequence of the connected domain pointing to the growth seed point of the connected domain are counted, and each of the vectors is recorded as the growth vector of each edge pixel point, and the arithmetic mean of the module lengths of all growth vectors is recorded as the overall growth distance of the connected domain.

6. The garden maintenance information processing system based on image processing according to claim 1 is characterized in that: The mutation conditions specifically refer to: A mutation threshold is preset. When the color-oriented consistency amount of the new connected domain is less than or equal to the mutation threshold, the mutation condition is met. When the color-oriented consistency amount of the new connected domain is greater than the mutation threshold, the mutation condition is not met.

7. The garden maintenance information processing system based on image processing according to claim 1 is characterized in that: The continuous guiding condition specifically refers to: The preceding continuity factor of the new connected domain is obtained, and the continuity guidance coefficient is preset. When the preceding continuity factor of the new connected domain is less than or equal to the continuity guidance coefficient, the continuity guidance condition is satisfied. When the preceding continuity factor of the new connected domain is greater than the continuity guidance coefficient, the continuity guidance condition is not satisfied.

8. The garden maintenance information processing system based on image processing according to claim 7 is characterized in that: The specific steps of obtaining the leading continuity factor of the new connected domain are as follows: Starting from the initial connected domain, the color-oriented consistency amount of the connected domain is stored in a set, and the set is recorded as a loop parameter set. Each time the color-oriented consistency amount of a new connected domain is calculated, the color-oriented consistency amount of the new connected domain is put into the loop parameter set and the arithmetic mean of all the color-oriented consistency amounts in the loop parameter set is calculated, and the arithmetic mean is recorded as the leading continuity factor of the new connected domain.

Citation Information

Patent Citations

  • Method for transferring colors by utilizing color space distribution

    CN102521807A

  • Image saliency target detection method based on k-means and level set superpixel segmentation

    CN110866896A

  • Junk image denoising processing method based on computer vision

    CN116152115A

  • Remote sensing image adaptive segmentation method based on land utilization type

    CN116188496A

  • Garden landscape greening area intelligent division method based on remote sensing image

    CN117291934A