WO2017215206A1 - Automatic plant identification system and method - Google Patents

Automatic plant identification system and method Download PDF

Info

Publication number
WO2017215206A1
WO2017215206A1 PCT/CN2016/108821 CN2016108821W WO2017215206A1 WO 2017215206 A1 WO2017215206 A1 WO 2017215206A1 CN 2016108821 W CN2016108821 W CN 2016108821W WO 2017215206 A1 WO2017215206 A1 WO 2017215206A1
Authority
WO
WIPO (PCT)
Prior art keywords
plant
information
image
module
identified
Prior art date
Application number
PCT/CN2016/108821
Other languages
French (fr)
Chinese (zh)
Inventor
张贯京
Original Assignee
深圳市易特科信息技术有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 深圳市易特科信息技术有限公司 filed Critical 深圳市易特科信息技术有限公司
Publication of WO2017215206A1 publication Critical patent/WO2017215206A1/en

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/40Extraction of image or video features
    • G06V10/44Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G06V10/443Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components by matching or filtering
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/20Image preprocessing
    • G06V10/30Noise filtering
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/40Extraction of image or video features
    • G06V10/56Extraction of image or video features relating to colour
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00Scenes; Scene-specific elements
    • G06V20/10Terrestrial scenes
    • G06V20/188Vegetation

Definitions

  • the present invention relates to the field of plant identification technology, and in particular, to an automatic plant identification system and method.
  • the main object of the present invention is to provide an automatic plant identification system and method, which aims to solve the problem that the plant non-professionals cannot quickly identify plants without knowing the name of the plant in the prior art, which is not conducive to plant non-professionals.
  • the present invention provides an automatic plant identification system.
  • the plant automatic identification system runs in a smart terminal, and the system includes a plant image information acquisition module
  • An image preprocessing module configured to acquire image information of the plant to be identified;
  • the image preprocessing module is configured to remove noise in image information of a plant to be identified
  • the plant feature extraction and matching module is configured to extract a plant wheel from image information subjected to denoising processing Profile characteristics and color features, and matching the extracted plant outline features and color features with the plant outline features and color feature information in the pre-established database, respectively, to obtain the matching degree between the plants to be identified and the existing plant information in the database;
  • the plant information output module is configured to output plant information with the greatest matching degree as information of the plant to be identified
  • the digitizing sub-module is configured to sample and quantize original image information of the plant to be identified into a digital image
  • the normalization sub-module is configured to retain a specific feature of the digital image by a transform process
  • the smoothing processing sub-module is configured to cancel noise in the digital image
  • an image complex atom module for correcting pixel degradation of the digital image
  • the image enhancement sub-module is configured to perform image enhancement processing on the information in the digital image to improve the visual effect of the digital image.
  • the plant feature extraction and matching module includes a contour feature extraction sub-module, a color feature extraction sub-module, and a matching degree calculation sub-module, wherein:
  • the contour feature extraction sub-module is configured to extract contour features in image information by using a Hu moment matching algorithm, a contour tree matching algorithm, or a pairwise geometric histogram matching algorithm;
  • the color feature extraction sub-module is configured to extract color features in the image information by using a color moment algorithm or a color set algorithm
  • the matching degree calculation sub-module is configured to match the contour feature and the color feature in the extracted image information with the plant outline feature and the color feature information in the pre-established database to obtain the plant to be identified and the database.
  • the degree of matching of plant information already exists.
  • the image information of the plant to be identified includes at least two kinds of information of leaves, neck, flowers and fruits of the plant to be identified.
  • the information of the plant to be identified includes at least the name, picture, growth habit and function of the plant.
  • the present invention also provides a method for automatically identifying plants.
  • the automatic plant identification method runs in a smart terminal, and the method includes the following steps: [0024] acquiring image information of the plant to be identified;
  • the plant information having the largest matching degree is output as the information of the plant to be identified.
  • the step of removing noise in the image information of the plant to be identified includes:
  • the plant outline feature and the color feature are extracted from the image information subjected to the denoising process, and the extracted plant outline feature and the color feature are respectively associated with the plant outline feature and the color feature information in the pre-established database.
  • the steps of matching to obtain the matching degree of the plant information to be recognized in the database and the database include:
  • contour feature and the color feature in the extracted image information are matched with the plant outline feature and the color feature information in the pre-established database to obtain the matching degree between the plant to be identified and the existing plant information in the database.
  • the image information of the plant to be identified includes at least two kinds of information of leaves, neck, flowers and fruits of the plant to be identified.
  • the information of the plant to be identified includes at least the name, picture, growth habit and function of the plant.
  • the automatic plant identification system and method provided by the present invention removes irrelevant information in an image by removing noise in image information of a plant to be identified, and recovers useful real information, and is preprocessed by extraction.
  • the contour features and color features of the image information are matched with the plant outline features and color feature information in the pre-established database to obtain the matching degree of the plant information in the database to be identified and the database, and the plant information with the largest matching degree is output.
  • the plant non-professionals can provide a quick way to identify plant information without knowing the name of the plant, which is beneficial to the survival of plant non-professionals in the wild.
  • FIG. 1 is a schematic diagram of functional modules of a preferred embodiment of the automatic plant identification system of the present invention
  • FIG. 2 is a schematic diagram of sub-function modules of an image pre-processing module in the automatic plant identification system of the present invention
  • FIG. 3 is a schematic diagram of a sub-function module of a plant feature extraction and matching module in the automatic plant identification system of the present invention
  • FIG. 4 is a schematic flow chart of a preferred embodiment of the automatic plant identification method of the present invention.
  • step S20 in FIG. 4 is a schematic flowchart of the refinement of step S20 in FIG. 4;
  • FIG. 6 is a schematic diagram showing the refinement process of step S30 in FIG. 4.
  • the present invention provides an automatic plant identification system and method for recovering useful real information by extracting noise in image information of a plant to be identified, eliminating irrelevant information in the image, by extracting
  • the contour features and color features of the preprocessed image information are matched with the plant outline features and color feature information in the pre-established database to obtain the matching degree of the plant information in the database to be identified and the database, and the output matching degree is the largest.
  • the plant information is for the user's reference for the information of the plant to be identified, and provides a quick identification plant letter for plant non-professionals without knowing the name of the plant. The way of interest is conducive to the survival of plant non-professionals in the wild.
  • FIG. 1 is a schematic diagram of functional modules of a preferred embodiment of the automatic plant identification system of the present invention.
  • the automatic plant identification system 10 runs in the smart terminal 1.
  • the smart terminal 1 is a microcomputer that is convenient to carry and has communication and information processing capabilities, such as a smart phone, an intelligent handheld terminal, or a smart tablet.
  • the intelligent terminal 1 further includes a storage unit 12, a processing unit 14, and a communication unit 16.
  • the storage unit 12 may be a read only storage unit ROM, an electrically erasable storage unit EEPRO M, a flash storage unit FLASH or a solid hard disk.
  • the processing unit 14 may be a central processing unit (CPU), a microcontroller (MCU), a data processing chip, or an information processing unit having data processing functions.
  • the communication unit 16 is a wireless communication interface or a wired interface with remote communication functions, for example, wireless or communication technologies supporting GSM, GPRS, WCDMA, CDMA, TD-SCDMA, WiMAX, TD-LTE, FDD-LTE, etc. Wired communication interface
  • the automatic plant identification system 10 includes a plant image information acquisition module 100, an image preprocessing module 102, a plant feature extraction and matching module 104, and a plant information output module 106.
  • the module referred to in the present invention refers to a series of computer program instruction segments that can be executed by the processing unit 14 and that are capable of performing fixed functions, which are stored in the storage unit 12.
  • the plant image information acquiring module 100 is configured to acquire image information of the plant to be identified, and the image information of the plant to be identified includes at least two kinds of information of leaves, necks, flowers, and fruits of the plants to be identified.
  • the information of the leaves includes at least the color and shape of the blade;
  • the information of the neck includes at least the color and diameter of the neck;
  • the information of the flower includes at least the shape and color of the flower;
  • the information of the fruit includes at least the color, shape and size of the fruit.
  • the image preprocessing module 102 is configured to remove noise in image information of the plant to be identified.
  • the main purpose of image preprocessing is to eliminate irrelevant information in the image, restore useful real information, enhance the detectability of the information and minimize the data, thereby improving the reliability of feature extraction, image segmentation, matching and recognition.
  • the image preprocessing process generally includes steps of digitization, normalization, smoothing, restoration, and enhancement.
  • FIG. 2 is a schematic diagram of a sub-function module of an image pre-processing module in the automatic plant identification system of the present invention.
  • the image preprocessing module 102 includes a digitization submodule 1020, a normalization submodule 1022, a smoothing submodule 1024, an image complex atom module 1026, and an image enhancement submodule.
  • Block 1028 Block 1028:
  • the digitizing sub-module 1020 is configured to sample and quantize the original image information of the plant to be identified to obtain a digital image that the processing unit 14 can process.
  • the gray value of the original image information (i.e., the original image) of a plant to be identified is a continuous function of a spatial variable (continuous value of position).
  • the original image is grayscaled and quantized on the MxN lattice (classified as one of 2b grayscales) to obtain a digital image that the computer can process.
  • the magnitudes of the M, N, and b values are preset.
  • the sampling period is equal to or less than half of the minimum detail period in the original image
  • the spectrum of the reconstructed image is equal to the spectrum of the original image, so the reconstructed image can be identical to the original image.
  • the product of M, N and b determines the storage amount of an image in the intelligent terminal
  • the appropriate M, N and b values are selected according to the different properties of the image under the condition of a certain amount of storage to obtain the best. Processing effect.
  • the normalization sub-module 1022 is configured to preserve a particular feature of the digital image by a transform process such that a particular feature of the digital image has a constant property under a given transform. Certain features of the digital image, such as the area and perimeter of the object, have constant properties for changes in coordinate rotation. In general, the influence of certain factors or transformations on some properties of the image can be eliminated or reduced by normalization and can be selected as the basis for measuring the image. For example, for remote sensing images with uncontrollable illumination, the normalization of grayscale histograms is necessary for image analysis. Gray Normalization, Geometric Normalization, and Transformation Normalization is the three normalization methods for obtaining the invariant properties of an image.
  • the smoothing sub-module 1024 is for canceling noise in the digital image.
  • Commonly used smoothing methods are median method, local averaging method and k-nearest neighbor averaging method.
  • the local area size can be fixed or it can be changed point by point with the gray value.
  • the image complex atom module 1026 is for correcting pixel degradation of the digital image such that the reconstructed or estimated digital image is as close as possible to the ideal digital image without degradation.
  • Image pixel degradation often occurs in practical applications. For example, disturbances in large airflow, aberrations in the optical system, and relative motion of the camera and objects can degrade the remote sensing image.
  • the basic restoration technique is to treat the acquired degraded image g(x, y) as a convolution of the degenerate function h(x, y) and the ideal image f(x, y). Their Fourier transform relationship
  • the image enhancement sub-module 1028 is configured to selectively perform image enhancement processing on information in the digital image to improve a visual effect of the digital image, or convert the digital image to be more suitable for the processing unit 14
  • the form of processing to facilitate data extraction or identification can highlight the outline of an image with a high-pass filter, allowing the machine to measure the shape and perimeter of the outline.
  • Contrast broadening, logarithmic transformation, density stratification, and histogram equalization can all be used to change image grays and highlight details.
  • the plant feature extraction and matching module 104 is configured to extract plant outline features and color features from the denoised processed image information, and extract the extracted plant outline features and color features respectively with plants in a pre-established database.
  • the contour feature and the color feature information are matched to obtain the matching degree between the plant to be identified and the existing plant information in the database.
  • FIG. 3 is a schematic diagram of sub-function modules of the plant feature extraction and matching module in the automatic plant identification system of the present invention.
  • the plant feature extraction and matching module 104 includes a contour feature extraction sub-module 1040, a color feature extraction sub-module 1042, and a matching degree calculation sub-module 1044.
  • the contour feature extraction sub-module 1040 is configured to extract contour features in the image information by using a Hu moment matching algorithm, a contour tree matching algorithm, or a pairwise geometric histogram matching algorithm.
  • the contour can be obtained by extracting and matching the contour features to obtain the shape, size, etc. of the elements of the plant image (for example: leaves, neck, flowers, fruits). Since the Hu moment of the contour can be invariant to changes including scaling, rotation, and mirror mapping, contour matching is performed through a common Hu moment matching algorithm.
  • the cvMatchShapes function makes it easy to match between two contours.
  • the contour tree matching algorithm compares two contours in the form of a tree.
  • the cvMa tchContourTrees function enables comparison of contour trees.
  • the pairwise geometric histogram matching algorithm refers to the use of histogram contrast method to perform contour matching after the paired geometric histograms of the contours are obtained.
  • the color feature extraction sub-module 1042 is configured to extract color features in the image information by using a color moment algorithm or a color set algorithm.
  • the color histogram is used to reflect the composition distribution of the image color, that is, the probability of occurrence of various colors.
  • the color moment algorithm uses the concept of moments in linear algebra to represent the color distribution in an image as a color moment. The color distribution is described by the color first moment (average Average), the color second moment (variance Variance), and the color third moment (skewness Skewness). Unlike color histograms, using color moments Line image descriptions do not require quantization of image features.
  • the color set algorithm implements retrieval of large-scale images based on color.
  • the method converts the color into the HSV color space
  • the image is divided into several regions according to the color information thereof, and the color is divided into multiple bins, and each region performs color space quantization to establish a color cable I, thereby establishing a binary image.
  • Color index table To speed up the search, you can also construct a binary search tree for feature retrieval.
  • the matching degree calculation sub-module 1044 is configured to match the contour feature and the color feature in the extracted image information with the plant outline feature and the color feature information in the pre-established database to obtain the plant and database to be identified.
  • the degree of matching of plant information already exists.
  • the pre-established database is run in a remote server, and the smart terminal 1 establishes a communication connection with the remote server through the communication unit 16, and reads pre-stored plant outline features and color feature information from the database of the remote server and matches them. .
  • the plant information output module 106 is configured to output plant information with the greatest matching degree as information of the plant to be identified.
  • the information of the plant to be identified includes at least the name, picture, growth habit and function of the plant for reference by the user, and provides a means for the plant non-professional to provide quick identification of plant information without knowing the name of the plant, which is beneficial to the plant non- Professionals survive in the wild.
  • the image matching and/or comparison process of the present invention is an image processing technology in the prior art, which is not limited and described herein.
  • the automatic plant identification system removes noise in the image information of the plant to be identified through the image preprocessing module, eliminates irrelevant information in the image, restores useful real information, and extracts through the plant feature extraction and matching module.
  • the contour feature and the color feature of the processed image information are matched with the plant outline feature and the color feature information in the pre-established database, and the matching degree of the plant information to be recognized in the database and the database is obtained, and the plant with the largest matching degree is outputted.
  • the information is for the reference of the information of the plants to be identified, and provides a way for plant non-professionals to quickly identify plant information without knowing the name of the plant, which is beneficial to the survival of plant non-professionals in the wild.
  • FIG. 4 is a schematic flow chart of a preferred embodiment of the automatic plant identification method of the present invention.
  • the automatic plant identification method runs on the smart In device 1, the method includes the following steps:
  • S10 acquiring image information of the plant to be identified
  • the plant image information acquiring module 100 acquires image information of the plant to be identified, and the image information of the plant to be identified includes at least two kinds of information of a leaf, a neck, a flower, and a fruit of the plant to be identified.
  • the information of the leaves includes at least the color and shape of the blade;
  • the information of the neck includes at least the color and diameter of the neck;
  • the information of the flower includes at least the shape and color of the flower;
  • the information of the fruit includes at least the color, shape and size of the fruit.
  • S20 removing noise in image information of the plant to be identified
  • the image pre-processing module 102 removes noise in the image information of the plant to be identified.
  • the main purpose of image preprocessing is to eliminate irrelevant information in the image, restore useful real information, enhance the detectability of the information and minimize the data, thereby improving the reliability of feature extraction, image segmentation, matching and recognition.
  • the image preprocessing process generally includes steps of digitization, normalization, smoothing, restoration, and enhancement.
  • FIG. 5 is a schematic diagram of the refinement process of step S20 in FIG. Step S20 includes the following steps
  • S201 Raw image information of the plant to be identified is sampled and quantized to obtain a digital image.
  • the digitizing sub-module 1020 samples and quantizes the original image information of the plant to be identified to obtain a digital image that the processing unit 14 can process.
  • the gray value of the original image information (i.e., the original image) of a plant to be identified is a continuous function of a spatial variable (continuous value of position).
  • the original image is grayscaled and quantized on the MxN lattice (classified as one of 2b grayscales) to obtain a digital image that the computer can process.
  • the magnitudes of the M, N, and b values are preset.
  • the sampling period is equal to or less than half of the minimum detail period in the original image
  • the spectrum of the reconstructed image is equal to the spectrum of the original image, so the reconstructed image can be identical to the original image.
  • the product of M, N and b determines the storage amount of an image in the intelligent terminal
  • the appropriate M, N and b values are selected according to the different properties of the image under the condition of a certain amount of storage to obtain the best. Processing effect.
  • S202 retaining a specific feature of the digital image by a transform process.
  • the normalization sub-module 1022 preserves certain features of the digital image by transform processing such that certain features of the digital image have invariant properties under a given transformation. Certain features of the image, such as the area and perimeter of the object, have invariant properties for coordinate rotation transformations. In general, the influence of certain factors or transformations on some properties of the image can be eliminated or attenuated by normalization, which can be selected as the basis for measuring the image. For example, for remote sensing images with uncontrollable illumination, the normalization of gray histograms is necessary for image analysis. Gray normalization, geometric normalization, and transform normalization are three normalization methods for obtaining the invariant properties of an image.
  • S203 Eliminate noise in the digital image.
  • the smoothing processing sub-module 1024 eliminates random noise in the digital image.
  • Commonly used smoothing methods are median method, local averaging method and k-nearest neighbor averaging method.
  • the size of the local area can be fixed or it can be changed point by point with the size of the gray value.
  • the image complex atom module 1026 corrects pixel degradation of the digital image caused by various causes, so that the reconstructed or estimated image is as close as possible to the ideal image field without degradation.
  • Image pixel degradation often occurs in practical applications. For example, disturbances in large airflow, aberrations in the optical system, and relative motion of the camera and objects can degrade the remote sensing image.
  • the basic restoration technique is to treat the acquired degraded image g(x, y) as a convolution of the degenerate function h(x, y) and the ideal image f(x, y).
  • F (u, v) After determining the degradation function according to the degradation mechanism, F (u, v) can be obtained from this relation. Using the inverse Fourier transform to find f(x, y), usually called the inverse filter. When there is no noise, the Wiener filter becomes the ideal inverse filter.
  • S205 Perform image enhancement processing on the information in the digital image to improve the visual effect of the digital image.
  • the image enhancement sub-module 1028 selectively enhances and suppresses information in the digital image to improve the visual effect of the digital image, or to transform the image into a form more suitable for processing by the processing unit 14.
  • an image enhancement system can highlight the outline of an image through a high-pass filter, enabling the machine to measure the shape and perimeter of the outline.
  • image enhancement techniques Contrast broadening, logarithmic transformation, density stratification, and histogram equalization can all be used to change the graph. Like gray tones and highlighting details.
  • S30 extracting plant outline features and color features from the image information after denoising processing, and matching the extracted plant outline features and color features with the plant outline features and color feature information in a pre-established database, respectively, The degree of matching between the plants to be identified and the existing plant information in the database is obtained.
  • FIG. 6 is a schematic flowchart of the refinement of step S30 in FIG. Step S30 includes the following steps:
  • S301 extracting contour features in the image information by using a Hu moment matching algorithm, a contour tree matching algorithm, or a pairwise geometric histogram matching algorithm.
  • the contour feature extraction sub-module extracts contour features in the image information by using a Hu moment matching algorithm, a contour tree matching algorithm, or a pairwise geometric histogram matching algorithm.
  • the contour can be extracted and matched by contour features to obtain the shape, size, etc. of the elements of the plant image (for example: leaves, neck, flowers, fruits). Since the Hu moment of the contour can be invariant to changes including scaling, rotation, and mirror mapping, contour matching is performed through a common Hu moment matching algorithm.
  • the cvMatchShapes function makes it easy to match between two contours.
  • the contour tree matching algorithm compares two contours in the form of a tree.
  • the cvMatchC ontourTrees function enables the comparison of contour trees.
  • the pairwise geometric histogram matching algorithm refers to the use of histogram contrast method to perform contour matching after obtaining the paired geometric histogram of the contour.
  • S302 Extracting color features in the image information by using a color moment algorithm or a color set algorithm.
  • the color feature extraction sub-module 1042 extracts color features in the image information using a color moment algorithm or a color set algorithm.
  • the color histogram is used to reflect the composition distribution of the image color, that is, the probability of occurrence of various colors.
  • the color moment algorithm uses the concept of moments in linear algebra to represent the color distribution in an image as a color moment. The color distribution is described by the color first moment (average Average), the color second moment (variance Variance), and the color third moment (skewness). Unlike color histograms, image description using color moments does not require quantization of image features.
  • the color set algorithm implements the retrieval of large-scale images based on color.
  • the method converts the color into the HSV color space
  • the image is divided into several regions according to the color information thereof, and the color is divided into multiple bins, and each region performs color space quantization to establish a color cable I, thereby establishing a binary image.
  • Color index table To speed up the search, you can also construct a binary search tree for feature retrieval.
  • S303 The contour feature and the color feature in the extracted image information are followed by a pre-established database.
  • the plant outline feature and the color feature information are matched to obtain the matching degree of the plant information to be identified and the existing plant information in the database.
  • the matching degree calculation sub-module 1044 matches the contour feature and the color feature in the extracted image information with the plant outline feature and the color feature information in the pre-established database to obtain the plant to be identified and the database.
  • the degree of matching of plant information already exists.
  • the pre-established database is run in a remote server, and the smart terminal 1 establishes a communication connection with the remote server through the communication unit 16, and reads pre-stored plant outline features and color feature information from the database of the remote server and matches them. .
  • S40 Output plant information with the greatest matching degree as information of the plant to be identified.
  • the plant information output module 106 outputs the plant information having the greatest matching degree as the information of the plant to be identified.
  • the information of the plant to be identified includes at least the name, picture, growth habit and function of the plant
  • the image matching and/or comparison process of the present invention is an image processing technology in the prior art, which is not limited and described herein.
  • the automatic plant identification method removes the noise in the image information of the plant to be identified, eliminates irrelevant information in the image, restores useful real information, and extracts contour features and color features of the preprocessed image information. And matching with the plant outline feature and color feature information in the pre-established database, obtaining the matching degree of the existing plant information in the database to be identified, and outputting the plant information with the largest matching degree as the information of the plant to be identified for the user's reference. It provides a way for plant non-professionals to quickly identify plant information without knowing the name of the plant, which is beneficial to the survival of plant non-professionals in the wild.
  • the automatic plant identification system and method provided by the present invention removes the plant to be identified Noise in the image information, eliminating irrelevant information in the image, restoring useful real information, by extracting contour features and color features of the preprocessed image information, and with plant contour features and color feature information in a pre-established database Matching is performed to obtain the matching degree of the plant information in the database to be identified, and the plant information with the largest matching degree is the information of the plant to be identified for the user's reference, and the plant non-professionals provide the fast without knowing the plant name.
  • the way to identify plant information is beneficial to the survival of plant non-professionals in the wild.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)

Abstract

An automatic plant identification system and method. The method comprises the following steps of: acquiring image information about a plant to be identified (S10); removing noise from the image information about the plant to be identified (S20); extracting a plant contour feature and a colour feature from the image information subjected to de-noising processing, and respectively matching the extracted plant contour feature information and colour feature with plant contour feature and colour feature information in a pre-established database, so as to obtain a degree of matching between the plant to be identified and the plant information existing in the database (S30); and outputting plant information with the maximum degree of matching as information about the plant to be identified (S40). By means of the method, a path for quickly identifying plant information is provided where a plant name is unknown to a plant non-specialist, wherein same facilitates the field survival of the plant non-specialist.

Description

植物自动识别系统和方法  Automatic plant identification system and method
技术领域  Technical field
[0001] 本发明涉及植物识别技术领域, 尤其涉及一种植物自动识别系统和方法。  [0001] The present invention relates to the field of plant identification technology, and in particular, to an automatic plant identification system and method.
背景技术  Background technique
[0002] 据估计, 地球上大约有 22万到 42万种不同类别的植物。 对于植物的分类识别是 一项庞大复杂的工作, 传统的植物识别方法主要依靠相应的植物学家, 利用他 们自身的专业知识, 对植物外形、 表皮、 叶子等进行研究分析, 确认植物类别 。 然而, 对于经常需要进行野外作战或工作的人员来说, 不具备专业的植物学 知识, 很难快速判断出植物所属分类以及名称, 更不能得知其生长生活习性。 实际上, 目前很多机构建立了很完备的电子化植物知识百科, 例如百度百科, 只要输入植物确切的名称即可査询到该植物的详细信, 但是在不知道植物名称 的情况下, 就无法获取到植物的详细信息。 因此, 目前现有技术主要的缺陷如 下: 植物非专业人员在不知道植物名称的情况下不能快速识别植物的信息, 不 利于其野外生存。  [0002] It is estimated that there are approximately 220,000 to 420,000 different types of plants on Earth. The classification and identification of plants is a huge and complex task. Traditional plant identification methods rely mainly on the corresponding botanists to use their own expertise to conduct research and analysis on plant shapes, skins, leaves, etc., to identify plant types. However, for those who often need to conduct field operations or work, without professional botany knowledge, it is difficult to quickly determine the classification and name of the plant, and not know its growth and habits. In fact, many institutions have established a very complete electronic plant knowledge encyclopedia, such as Baidu Encyclopedia, just enter the exact name of the plant to find the detailed information of the plant, but without knowing the name of the plant, it can not Get detailed information about the plant. Therefore, the main shortcomings of the prior art are as follows: Plant non-professionals cannot quickly identify plant information without knowing the name of the plant, which is not conducive to its survival in the wild.
技术问题  technical problem
[0003] 本发明的主要目的在于提供一种植物自动识别系统和方法, 旨在解决现有技术 中植物非专业人员在不知道植物名称的情况下无法快速识别植物的信息, 不利 于植物非专业人员野外生存的技术问题。  [0003] The main object of the present invention is to provide an automatic plant identification system and method, which aims to solve the problem that the plant non-professionals cannot quickly identify plants without knowing the name of the plant in the prior art, which is not conducive to plant non-professionals. Technical problems of personnel survival in the wild.
问题的解决方案  Problem solution
技术解决方案  Technical solution
[0004] 为实现上述目的, 本发明提供了一种植物自动识别系统。  In order to achieve the above object, the present invention provides an automatic plant identification system.
[0005] 所述植物自动识别系统运行于智能终端中, 该系统包括植物图像信息获取模块 [0005] The plant automatic identification system runs in a smart terminal, and the system includes a plant image information acquisition module
、 图像预处理模块、 植物特征提取和匹配模块以及植物信息输出模块, 其中: [0006] 所述植物图像信息获取模块用于获取待识别植物的图像信息; An image preprocessing module, a plant feature extraction and matching module, and a plant information output module, wherein: the plant image information acquisition module is configured to acquire image information of the plant to be identified;
[0007] 所述图像预处理模块用于去除待识别植物的图像信息中的噪声; [0007] the image preprocessing module is configured to remove noise in image information of a plant to be identified;
[0008] 所述植物特征提取和匹配模块用于从经过去噪声处理的图像信息中提取植物轮 廓特征和颜色特征, 并将提取的植物轮廓特征和颜色特征分别与预先建立的数 据库中的植物轮廓特征和颜色特征信息进行匹配, 得到待识别植物与数据库中 已存在植物信息的匹配度; [0008] The plant feature extraction and matching module is configured to extract a plant wheel from image information subjected to denoising processing Profile characteristics and color features, and matching the extracted plant outline features and color features with the plant outline features and color feature information in the pre-established database, respectively, to obtain the matching degree between the plants to be identified and the existing plant information in the database;
[0009] 所述植物信息输出模块用于输出匹配度最大的植物信息作为待识别植物的信息  [0009] the plant information output module is configured to output plant information with the greatest matching degree as information of the plant to be identified
[0010] 进一步地, 所述数字化子模块用于将待识别植物的原始图像信息采样并量化得 到数字图像; [0010] Further, the digitizing sub-module is configured to sample and quantize original image information of the plant to be identified into a digital image;
[0011] 所述归一化子模块用于通过变换处理保留所述数字图像的特定特征;  [0011] the normalization sub-module is configured to retain a specific feature of the digital image by a transform process;
[0012] 所述平滑处理子模块用于消除所述数字图像中的噪声; [0012] the smoothing processing sub-module is configured to cancel noise in the digital image;
[0013] 图像复原子模块用于校正所述数字图像的像素退化; [0013] an image complex atom module for correcting pixel degradation of the digital image;
[0014] 图像增强子模块用于对所述数字图像中的信息进行图像增强处理以改善所述数 字图像的视觉效果。  [0014] The image enhancement sub-module is configured to perform image enhancement processing on the information in the digital image to improve the visual effect of the digital image.
[0015] 进一步地, 植物特征提取和匹配模块包括轮廓特征提取子模块、 颜色特征提取 子模块以及匹配度计算子模块, 其中:  [0015] Further, the plant feature extraction and matching module includes a contour feature extraction sub-module, a color feature extraction sub-module, and a matching degree calculation sub-module, wherein:
[0016] 所述轮廓特征提取子模块用于采用 Hu矩匹配算法、 轮廓树匹配算法或成对几何 直方图匹配算法提取图像信息中的轮廓特征; [0016] the contour feature extraction sub-module is configured to extract contour features in image information by using a Hu moment matching algorithm, a contour tree matching algorithm, or a pairwise geometric histogram matching algorithm;
[0017] 所述颜色特征提取子模块用于采用颜色矩算法或颜色集算法提取图像信息中的 颜色特征; [0017] the color feature extraction sub-module is configured to extract color features in the image information by using a color moment algorithm or a color set algorithm;
[0018] 所述匹配度计算子模块用于将提取到的图像信息中的轮廓特征和颜色特征后与 预先建立的数据库中的植物轮廓特征和颜色特征信息进行匹配, 得到待识别植 物与数据库中已存在植物信息的匹配度。  [0018] The matching degree calculation sub-module is configured to match the contour feature and the color feature in the extracted image information with the plant outline feature and the color feature information in the pre-established database to obtain the plant to be identified and the database. The degree of matching of plant information already exists.
[0019] 优选地, 所述待识别植物的图像信息包括待识别植物的叶子、 颈、 花和果实中 的至少两种信息。  [0019] Preferably, the image information of the plant to be identified includes at least two kinds of information of leaves, neck, flowers and fruits of the plant to be identified.
[0020] 优选地, 所述待识别植物的信息至少包括该植物的名称、 图片、 生长习性以及 功能。  [0020] Preferably, the information of the plant to be identified includes at least the name, picture, growth habit and function of the plant.
[0021]  [0021]
[0022] 本发明还提供了一种植物自动识别方法。  [0022] The present invention also provides a method for automatically identifying plants.
[0023] 所述植物自动识别方法运行于智能终端中, 该方法包括如下步骤: [0024] 获取待识别植物的图像信息; [0023] The automatic plant identification method runs in a smart terminal, and the method includes the following steps: [0024] acquiring image information of the plant to be identified;
[0025] 去除待识别植物的图像信息中的噪声;  [0025] removing noise in image information of the plant to be identified;
[0026] 从经过去噪声处理的图像信息中提取植物轮廓特征和颜色特征, 并将提取的植 物轮廓特征和颜色特征分别与预先建立的数据库中的植物轮廓特征和颜色特征 信息进行匹配, 得到待识别植物与数据库中已存在植物信息的匹配度;  Extracting plant outline features and color features from the image information subjected to denoising processing, and matching the extracted plant outline features and color features with plant outline features and color feature information in a pre-established database, respectively, to be obtained Identify the degree of match between the plant and the existing plant information in the database;
[0027] 输出匹配度最大的植物信息作为待识别植物的信息。  [0027] The plant information having the largest matching degree is output as the information of the plant to be identified.
[0028] 进一步地, 所述去除待识别植物的图像信息中的噪声的步骤包括:  [0028] Further, the step of removing noise in the image information of the plant to be identified includes:
[0029] 将待识别植物的原始图像信息采样并量化得到数字图像;  [0029] sampling and quantizing the original image information of the plant to be identified to obtain a digital image;
[0030] 通过变换处理保留所述数字图像的特定特征;  [0030] retaining specific features of the digital image by transform processing;
[0031] 消除所述数字图像中的噪声;  [0031] eliminating noise in the digital image;
[0032] 校正所述数字图像的像素退化;  Correcting pixel degradation of the digital image;
[0033] 对所述数字图像中的信息进行图像增强处理以改善所述数字图像的视觉效果。  [0033] performing image enhancement processing on the information in the digital image to improve the visual effect of the digital image.
[0034] 进一步地, 所述从经过去噪声处理的图像信息中提取植物轮廓特征和颜色特征 , 并将提取的植物轮廓特征和颜色特征分别与预先建立的数据库中的植物轮廓 特征和颜色特征信息进行匹配, 得到待识别植物与数据库中已存在植物信息的 匹配度的步骤包括: [0034] Further, the plant outline feature and the color feature are extracted from the image information subjected to the denoising process, and the extracted plant outline feature and the color feature are respectively associated with the plant outline feature and the color feature information in the pre-established database. The steps of matching to obtain the matching degree of the plant information to be recognized in the database and the database include:
[0035] 采用 Hu矩匹配算法、 轮廓树匹配算法或成对几何直方图匹配算法提取图像信息 中的轮廓特征;  [0035] extracting contour features in the image information by using a Hu moment matching algorithm, a contour tree matching algorithm, or a pairwise geometric histogram matching algorithm;
[0036] 采用颜色矩算法或颜色集算法提取图像信息中的颜色特征;  [0036] extracting color features in the image information by using a color moment algorithm or a color set algorithm;
[0037] 将提取到的图像信息中的轮廓特征和颜色特征后与预先建立的数据库中的植物 轮廓特征和颜色特征信息进行匹配, 得到待识别植物与数据库中已存在植物信 息的匹配度。  [0037] The contour feature and the color feature in the extracted image information are matched with the plant outline feature and the color feature information in the pre-established database to obtain the matching degree between the plant to be identified and the existing plant information in the database.
[0038] 优选地, 所述待识别植物的图像信息包括待识别植物的叶子、 颈、 花和果实中 的至少两种信息。  [0038] Preferably, the image information of the plant to be identified includes at least two kinds of information of leaves, neck, flowers and fruits of the plant to be identified.
[0039] 优选地, 所述待识别植物的信息至少包括该植物的名称、 图片、 生长习性以及 功能。  [0039] Preferably, the information of the plant to be identified includes at least the name, picture, growth habit and function of the plant.
发明的有益效果  Advantageous effects of the invention
有益效果 [0040] 相较于现有技术, 本发明提供的植物自动识别系统和方法通过去除待识别植物 的图像信息中的噪声, 消除图像中无关的信息, 恢复有用的真实信息, 通过提 取经过预处理的图像信息的轮廓特征和颜色特征, 并与预先建立的数据库中的 植物轮廓特征和颜色特征信息进行匹配, 得到待识别植物与数据库中已存在植 物信息的匹配度, 输出匹配度最大的植物信息为待识别植物的信息供用户参考 , 为植物非专业人员在不知道植物名称的情况下提供快速识别植物信息的途径 , 有利于植物非专业人员野外生存。 Beneficial effect Compared with the prior art, the automatic plant identification system and method provided by the present invention removes irrelevant information in an image by removing noise in image information of a plant to be identified, and recovers useful real information, and is preprocessed by extraction. The contour features and color features of the image information are matched with the plant outline features and color feature information in the pre-established database to obtain the matching degree of the plant information in the database to be identified and the database, and the plant information with the largest matching degree is output. For the information of the plants to be identified for the user's reference, the plant non-professionals can provide a quick way to identify plant information without knowing the name of the plant, which is beneficial to the survival of plant non-professionals in the wild.
对附图的简要说明  Brief description of the drawing
附图说明  DRAWINGS
[0041] 图 1为本发明植物自动识别系统较佳实施例的功能模块示意图;  1 is a schematic diagram of functional modules of a preferred embodiment of the automatic plant identification system of the present invention;
[0042] 图 2为本发明植物自动识别系统中图像预处理模块的子功能模块示意图; 2 is a schematic diagram of sub-function modules of an image pre-processing module in the automatic plant identification system of the present invention;
[0043] 图 3为本发明植物自动识别系统中植物特征提取和匹配模块的子功能模块示意 图; 3 is a schematic diagram of a sub-function module of a plant feature extraction and matching module in the automatic plant identification system of the present invention;
[0044] 图 4为本发明植物自动识别方法较佳实施例的流程示意图;  4 is a schematic flow chart of a preferred embodiment of the automatic plant identification method of the present invention;
[0045] 图 5为图 4中步骤 S20的细化流程示意图; 5 is a schematic flowchart of the refinement of step S20 in FIG. 4;
[0046] 图 6为图 4中步骤 S30的细化流程示意图。 6 is a schematic diagram showing the refinement process of step S30 in FIG. 4.
实施该发明的最佳实施例  BEST MODE FOR CARRYING OUT THE INVENTION
本发明的最佳实施方式  BEST MODE FOR CARRYING OUT THE INVENTION
[0047] 为更进一步阐述本发明为达成上述目的所采取的技术手段及功效, 以下结合附 图及较佳实施例, 对本发明的具体实施方式、 结构、 特征及其功效进行详细说 明。 应当理解, 此处所描述的具体实施例仅仅用以解释本发明, 并不用于限定 本发明。 The specific embodiments, structures, features and functions of the present invention are described in detail below with reference to the accompanying drawings and preferred embodiments. It is understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
[0048] 为实现本发明目的, 本发明提供了一种植物自动识别系统和方法, 通过去除待 识别植物的图像信息中的噪声, 消除图像中无关的信息, 恢复有用的真实信息 , 通过提取经过预处理的图像信息的轮廓特征和颜色特征, 并与预先建立的数 据库中的植物轮廓特征和颜色特征信息进行匹配, 得到待识别植物与数据库中 已存在植物信息的匹配度, 输出匹配度最大的植物信息为待识别植物的信息供 用户参考, 为植物非专业人员在不知道植物名称的情况下提供快速识别植物信 息的途径, 有利于植物非专业人员野外生存。 [0048] In order to achieve the object of the present invention, the present invention provides an automatic plant identification system and method for recovering useful real information by extracting noise in image information of a plant to be identified, eliminating irrelevant information in the image, by extracting The contour features and color features of the preprocessed image information are matched with the plant outline features and color feature information in the pre-established database to obtain the matching degree of the plant information in the database to be identified and the database, and the output matching degree is the largest. The plant information is for the user's reference for the information of the plant to be identified, and provides a quick identification plant letter for plant non-professionals without knowing the name of the plant. The way of interest is conducive to the survival of plant non-professionals in the wild.
[0049] 参照图 1所示, 图 1为本发明植物自动识别系统较佳实施例的功能模块示意图。 Referring to FIG. 1, FIG. 1 is a schematic diagram of functional modules of a preferred embodiment of the automatic plant identification system of the present invention.
[0050] 本发明提供的植物自动识别系统 10运行于智能终端 1中, 所述智能终端 1为智能 手机、 智能手持终端或智能平板电脑等携带方便且具有通讯和信息处理能力的 微型计算机。 所述智能终端 1还包括存储单元 12、 处理单元 14和通讯单元 16。 The automatic plant identification system 10 provided by the present invention runs in the smart terminal 1. The smart terminal 1 is a microcomputer that is convenient to carry and has communication and information processing capabilities, such as a smart phone, an intelligent handheld terminal, or a smart tablet. The intelligent terminal 1 further includes a storage unit 12, a processing unit 14, and a communication unit 16.
[0051] 所述的存储单元 12可以为一种只读存储单元 ROM, 电可擦写存储单元 EEPRO M、 快闪存储单元 FLASH或固体硬盘等。 所述的处理单元 14可以为一种中央处 理器 (Central Processing Unit, CPU) 、 微控制器 (MCU) 、 数据处理芯片、 或 者具有数据处理功能的信息处理单元。 所述通讯单元 16为一种具有远程通讯功 能的无线通讯接口或有线接口, 例如, 支持 GSM、 GPRS、 WCDMA、 CDMA、 TD-SCDMA、 WiMAX、 TD-LTE、 FDD-LTE等通讯技术的无线或有线通讯接口 [0051] The storage unit 12 may be a read only storage unit ROM, an electrically erasable storage unit EEPRO M, a flash storage unit FLASH or a solid hard disk. The processing unit 14 may be a central processing unit (CPU), a microcontroller (MCU), a data processing chip, or an information processing unit having data processing functions. The communication unit 16 is a wireless communication interface or a wired interface with remote communication functions, for example, wireless or communication technologies supporting GSM, GPRS, WCDMA, CDMA, TD-SCDMA, WiMAX, TD-LTE, FDD-LTE, etc. Wired communication interface
[0052] 在本实施例中, 植物自动识别系统 10包括植物图像信息获取模块 100、 图像预 处理模块 102、 植物特征提取和匹配模块 104以及植物信息输出模块 106。 本发明 所称的模块是指一种能够被所述处理单元 14执行并且能够完成固定功能的一系 列计算机程序指令段, 其存储在存储单元 12中。 In the present embodiment, the automatic plant identification system 10 includes a plant image information acquisition module 100, an image preprocessing module 102, a plant feature extraction and matching module 104, and a plant information output module 106. The module referred to in the present invention refers to a series of computer program instruction segments that can be executed by the processing unit 14 and that are capable of performing fixed functions, which are stored in the storage unit 12.
[0053] 所述植物图像信息获取模块 100用于获取待识别植物的图像信息, 所述待识别 植物的图像信息包括待识别植物的叶子、 颈、 花和果实中的至少两种信息。 其 中叶子的信息至少包括叶片颜色和形状; 颈的信息至少包括颈部颜色和直径参 数; 花的信息至少包括花的形状和颜色; 果实的信息至少包括果实的颜色、 形 状和大小。  [0053] The plant image information acquiring module 100 is configured to acquire image information of the plant to be identified, and the image information of the plant to be identified includes at least two kinds of information of leaves, necks, flowers, and fruits of the plants to be identified. The information of the leaves includes at least the color and shape of the blade; the information of the neck includes at least the color and diameter of the neck; the information of the flower includes at least the shape and color of the flower; and the information of the fruit includes at least the color, shape and size of the fruit.
[0054] 所述图像预处理模块 102用于去除待识别植物的图像信息中的噪声。 图像预处 理的主要目的是消除图像中无关的信息, 恢复有用的真实信息, 增强有关信息 的可检测性和最大限度地简化数据, 从而改进特征抽取、 图像分割、 匹配和识 别的可靠性。 图像预处理过程一般包括数字化、 归一化、 平滑、 复原和增强等 步骤。 具体地, 参照图 2所示, 图 2为本发明植物自动识别系统中图像预处理模 块的子功能模块示意图。 所述图像预处理模块 102包括数字化子模块 1020、 归一 化子模块 1022、 平滑处理子模块 1024、 图像复原子模块 1026以及图像增强子模 块 1028: [0054] The image preprocessing module 102 is configured to remove noise in image information of the plant to be identified. The main purpose of image preprocessing is to eliminate irrelevant information in the image, restore useful real information, enhance the detectability of the information and minimize the data, thereby improving the reliability of feature extraction, image segmentation, matching and recognition. The image preprocessing process generally includes steps of digitization, normalization, smoothing, restoration, and enhancement. Specifically, referring to FIG. 2, FIG. 2 is a schematic diagram of a sub-function module of an image pre-processing module in the automatic plant identification system of the present invention. The image preprocessing module 102 includes a digitization submodule 1020, a normalization submodule 1022, a smoothing submodule 1024, an image complex atom module 1026, and an image enhancement submodule. Block 1028:
[0055] 所述数字化子模块 1020用于将待识别植物的原始图像信息采样并量化得到处理 单元 14能够处理的数字图像。 具体地, 一幅待识别植物的原始图像信息 (即原 始图像) 的灰度值是空间变量 (位置的连续值) 的连续函数。 在 MxN点阵上对 该原始图像灰度采样并加以量化 (归为 2b个灰度等级之一) , 可以得到计算机 能够处理的数字图像。 为了使数字图像能重建原来的图像, 对M、 N和 b值的大 小预先设置。 在接收装置的空间和灰度分辨能力范围内, M、 N和 b的数值越大 , 重建图像的质量就越好。 当取样周期等于或小于原始图像中最小细节周期的 一半吋, 重建图像的频谱等于原始图像的频谱, 因此重建图像与原始图像可以 完全相同。 由于 M、 N和 b三者的乘积决定一幅图像在智能终端中的存储量, 因 此在存储量一定的条件下根据图像的不同性质选择合适的M、 N和 b值, 以获取 最好的处理效果。  [0055] The digitizing sub-module 1020 is configured to sample and quantize the original image information of the plant to be identified to obtain a digital image that the processing unit 14 can process. Specifically, the gray value of the original image information (i.e., the original image) of a plant to be identified is a continuous function of a spatial variable (continuous value of position). The original image is grayscaled and quantized on the MxN lattice (classified as one of 2b grayscales) to obtain a digital image that the computer can process. In order to enable the digital image to reconstruct the original image, the magnitudes of the M, N, and b values are preset. Within the spatial and grayscale resolution capabilities of the receiving device, the larger the values of M, N, and b, the better the quality of the reconstructed image. When the sampling period is equal to or less than half of the minimum detail period in the original image, the spectrum of the reconstructed image is equal to the spectrum of the original image, so the reconstructed image can be identical to the original image. Since the product of M, N and b determines the storage amount of an image in the intelligent terminal, the appropriate M, N and b values are selected according to the different properties of the image under the condition of a certain amount of storage to obtain the best. Processing effect.
[0056] 所述归一化子模块 1022用于通过变换处理保留所述数字图像的特定特征, 使所 述数字图像的特定特征在给定变换下具有不变的性质。 数字图像的特定特征, 例如物体的面积和周长, 对于坐标旋转变化来说就具有不变的性质。 在一般情 况下, 某些因素或变换对图像一些性质的影响可通过归一化处理得到消除或减 弱, 从而可以被选作测量图像的依据。 例如对于光照不可控的遥感图片, 灰度 直方图的归一化对于图像分析是十分必要的。 灰度归一化、 几何归一化和变换 归一化是获取图像不变性质的三种归一化方法。  [0056] The normalization sub-module 1022 is configured to preserve a particular feature of the digital image by a transform process such that a particular feature of the digital image has a constant property under a given transform. Certain features of the digital image, such as the area and perimeter of the object, have constant properties for changes in coordinate rotation. In general, the influence of certain factors or transformations on some properties of the image can be eliminated or reduced by normalization and can be selected as the basis for measuring the image. For example, for remote sensing images with uncontrollable illumination, the normalization of grayscale histograms is necessary for image analysis. Gray Normalization, Geometric Normalization, and Transformation Normalization is the three normalization methods for obtaining the invariant properties of an image.
[0057] 所述平滑处理子模块 1024用于消除所述数字图像中的噪声。 常用的平滑方法有 中值法、 局部求平均法和 k近邻平均法。 局部区域大小可以是固定的, 也可以是 逐点随灰度值大小变化的。 此外, 有吋应用空间频率域带通滤波方法。  [0057] The smoothing sub-module 1024 is for canceling noise in the digital image. Commonly used smoothing methods are median method, local averaging method and k-nearest neighbor averaging method. The local area size can be fixed or it can be changed point by point with the gray value. In addition, there is a method of applying a spatial frequency domain bandpass filtering method.
[0058] 图像复原子模块 1026用于校正所述数字图像的像素退化, 使重建或估计得到的 数字图像尽可能逼近于理想无退化的数字图像。 在实际应用中常常发生图像像 素退化现象。 例如大气流的扰动, 光学系统的像差, 相机和物体的相对运动都 会使遥感图像发生退化。 基本的复原技术是把获取的退化图像 g(x, y)看成是退 化函数 h(x, y)和理想图像 f(x, y)的卷积。 它们的傅里叶变换存在关系  [0058] The image complex atom module 1026 is for correcting pixel degradation of the digital image such that the reconstructed or estimated digital image is as close as possible to the ideal digital image without degradation. Image pixel degradation often occurs in practical applications. For example, disturbances in large airflow, aberrations in the optical system, and relative motion of the camera and objects can degrade the remote sensing image. The basic restoration technique is to treat the acquired degraded image g(x, y) as a convolution of the degenerate function h(x, y) and the ideal image f(x, y). Their Fourier transform relationship
G(u, v=H(u, v)F(u, v)。 根据退化机理确定退化函数后, 就可从此关系式求出 F (u, v), 再用傅里叶反变换求出 f(x, y), 通常称为反向滤波器。 当不存在噪声吋 , 维纳滤波器成为理想的反向滤波器。 G(u, v=H(u, v)F(u, v). After determining the degradation function according to the degradation mechanism, we can find F from this relation. (u, v), then find the f(x, y) by the inverse Fourier transform, which is usually called the inverse filter. When there is no noise, the Wiener filter becomes the ideal inverse filter.
[0059] 图像增强子模块 1028用于对所述数字图像中的信息有选择地进行图像增强处理 以改善所述数字图像的视觉效果, 或将所述数字图像转变为更适合于处理单元 1 4处理的形式, 以便于数据抽取或识别。 例如一个图像增强系统可以通过高通滤 波器来突出图像的轮廓线, 从而使机器能够测量轮廓线的形状和周长。 图像增 强技术有多种方法, 反差展宽、 对数变换、 密度分层和直方图均衡等都可用于 改变图像灰调和突出细节。  [0059] The image enhancement sub-module 1028 is configured to selectively perform image enhancement processing on information in the digital image to improve a visual effect of the digital image, or convert the digital image to be more suitable for the processing unit 14 The form of processing to facilitate data extraction or identification. For example, an image enhancement system can highlight the outline of an image with a high-pass filter, allowing the machine to measure the shape and perimeter of the outline. There are several ways to enhance image enhancement. Contrast broadening, logarithmic transformation, density stratification, and histogram equalization can all be used to change image grays and highlight details.
[0060] 所述植物特征提取和匹配模块 104用于从经过去噪声处理的图像信息中提取植 物轮廓特征和颜色特征, 并将提取的植物轮廓特征和颜色特征分别与预先建立 的数据库中的植物轮廓特征和颜色特征信息进行匹配, 得到待识别植物与数据 库中已存在植物信息的匹配度。 具体地, 参照图 3所示, 图 3为本发明植物自动 识别系统中植物特征提取和匹配模块的子功能模块示意图。 该植物特征提取和 匹配模块 104包括轮廓特征提取子模块 1040、 颜色特征提取子模块 1042以及匹配 度计算子模块 1044。  [0060] The plant feature extraction and matching module 104 is configured to extract plant outline features and color features from the denoised processed image information, and extract the extracted plant outline features and color features respectively with plants in a pre-established database. The contour feature and the color feature information are matched to obtain the matching degree between the plant to be identified and the existing plant information in the database. Specifically, referring to FIG. 3, FIG. 3 is a schematic diagram of sub-function modules of the plant feature extraction and matching module in the automatic plant identification system of the present invention. The plant feature extraction and matching module 104 includes a contour feature extraction sub-module 1040, a color feature extraction sub-module 1042, and a matching degree calculation sub-module 1044.
[0061] 所述轮廓特征提取子模块 1040用于采用 Hu矩匹配算法、 轮廓树匹配算法或成对 几何直方图匹配算法提取图像信息中的轮廓特征。 轮廓通过轮廓特征的提取和 匹配能够获得植物图像的要素 (例如: 叶子、 颈、 花、 果实) 的形状、 大小等 。 由于轮廓的 Hu矩能够对包括缩放、 旋转和镜像映射在内的变化具有不变性, 因此在进行轮廓匹配吋通过常用 Hu矩匹配算法。 cvMatchShapes函数可以很方便 的实现对 2个轮廓间的匹配。 轮廓树匹配算法采用树的形式比较两个轮廓。 cvMa tchContourTrees函数能够实现轮廓树的对比。 成对几何直方图匹配算法是指在得 到轮廓的成对几何直方图之后, 使用直方图对比的方法来进行轮廓匹配。  [0061] The contour feature extraction sub-module 1040 is configured to extract contour features in the image information by using a Hu moment matching algorithm, a contour tree matching algorithm, or a pairwise geometric histogram matching algorithm. The contour can be obtained by extracting and matching the contour features to obtain the shape, size, etc. of the elements of the plant image (for example: leaves, neck, flowers, fruits). Since the Hu moment of the contour can be invariant to changes including scaling, rotation, and mirror mapping, contour matching is performed through a common Hu moment matching algorithm. The cvMatchShapes function makes it easy to match between two contours. The contour tree matching algorithm compares two contours in the form of a tree. The cvMa tchContourTrees function enables comparison of contour trees. The pairwise geometric histogram matching algorithm refers to the use of histogram contrast method to perform contour matching after the paired geometric histograms of the contours are obtained.
[0062] 所述颜色特征提取子模块 1042用于采用颜色矩算法或颜色集算法提取图像信息 中的颜色特征。 颜色直方图用以反映图像颜色的组成分布, 即各种颜色出现的 概率。 颜色矩算法利用线性代数中矩的概念, 将图像中的颜色分布用颜色矩表 示。 利用颜色一阶矩 (平均值 Average) 、 颜色二阶矩 (方差 Variance) 和颜色 三阶矩 (偏斜度 Skewness) 来描述颜色分布。 与颜色直方图不同, 利用颜色矩进 行图像描述无需量化图像特征。 颜色集算法基于颜色实现对大规模图像的检索 。 该方法将颜色转化到 HSV颜色空间后, 将图像根据其颜色信息进行图像分割 成若干 region, 并将颜色分为多个 bin, 每个 region进行颜色空间量化建立颜色索 弓 I, 进而建立二进制图像颜色索引表。 为加快査找速度, 还可以构造二分査找 树进行特征检索。 [0062] The color feature extraction sub-module 1042 is configured to extract color features in the image information by using a color moment algorithm or a color set algorithm. The color histogram is used to reflect the composition distribution of the image color, that is, the probability of occurrence of various colors. The color moment algorithm uses the concept of moments in linear algebra to represent the color distribution in an image as a color moment. The color distribution is described by the color first moment (average Average), the color second moment (variance Variance), and the color third moment (skewness Skewness). Unlike color histograms, using color moments Line image descriptions do not require quantization of image features. The color set algorithm implements retrieval of large-scale images based on color. After the method converts the color into the HSV color space, the image is divided into several regions according to the color information thereof, and the color is divided into multiple bins, and each region performs color space quantization to establish a color cable I, thereby establishing a binary image. Color index table. To speed up the search, you can also construct a binary search tree for feature retrieval.
[0063] 所述匹配度计算子模块 1044用于将提取到的图像信息中的轮廓特征和颜色特征 后与预先建立的数据库中的植物轮廓特征和颜色特征信息进行匹配, 得到待识 别植物与数据库中已存在植物信息的匹配度。 具体地, 所述预先建立的数据库 运行于远程服务器中, 智能终端 1通过通讯单元 16与远程服务器建立通讯连接, 从远程服务器的数据库中读取预先存储的植物轮廓特征和颜色特征信息并进行 匹配。  [0063] The matching degree calculation sub-module 1044 is configured to match the contour feature and the color feature in the extracted image information with the plant outline feature and the color feature information in the pre-established database to obtain the plant and database to be identified. The degree of matching of plant information already exists. Specifically, the pre-established database is run in a remote server, and the smart terminal 1 establishes a communication connection with the remote server through the communication unit 16, and reads pre-stored plant outline features and color feature information from the database of the remote server and matches them. .
[0064] 所述植物信息输出模块 106用于输出匹配度最大的植物信息作为待识别植物的 信息。 该待识别植物的信息至少包括该植物的名称、 图片、 生长习性以及功能 , 以供用户参考, 为植物非专业人员在不知道植物名称的情况下提供快速识别 植物信息的途径, 有利于植物非专业人员野外生存。  [0064] The plant information output module 106 is configured to output plant information with the greatest matching degree as information of the plant to be identified. The information of the plant to be identified includes at least the name, picture, growth habit and function of the plant for reference by the user, and provides a means for the plant non-professional to provide quick identification of plant information without knowing the name of the plant, which is beneficial to the plant non- Professionals survive in the wild.
[0065] 本发明所述图像匹配和 /或比对过程为现有技术中的图像处理技术, 在此不做 限定和赘述。  The image matching and/or comparison process of the present invention is an image processing technology in the prior art, which is not limited and described herein.
[0066] 本发明提供的植物自动识别系统通过图像预处理模块去除待识别植物的图像信 息中的噪声, 消除图像中无关的信息, 恢复有用的真实信息, 通过植物特征提 取和匹配模块提取经过预处理的图像信息的轮廓特征和颜色特征, 并与预先建 立的数据库中的植物轮廓特征和颜色特征信息进行匹配, 得到待识别植物与数 据库中已存在植物信息的匹配度, 输出匹配度最大的植物信息为待识别植物的 信息供用户参考, 为植物非专业人员在不知道植物名称的情况下提供快速识别 植物信息的途径, 有利于植物非专业人员野外生存。  [0066] The automatic plant identification system provided by the invention removes noise in the image information of the plant to be identified through the image preprocessing module, eliminates irrelevant information in the image, restores useful real information, and extracts through the plant feature extraction and matching module. The contour feature and the color feature of the processed image information are matched with the plant outline feature and the color feature information in the pre-established database, and the matching degree of the plant information to be recognized in the database and the database is obtained, and the plant with the largest matching degree is outputted. The information is for the reference of the information of the plants to be identified, and provides a way for plant non-professionals to quickly identify plant information without knowing the name of the plant, which is beneficial to the survival of plant non-professionals in the wild.
[0067]  [0067]
[0068] 本发明的另外一个方面, 提供了一种与上述植物自动识别系统对应的方法。  In another aspect of the invention, a method corresponding to the plant automatic identification system described above is provided.
[0069] 参照图 4所示, 图 4为本发明植物自动识别方法较佳实施例的流程示意图。 [0069] Referring to FIG. 4, FIG. 4 is a schematic flow chart of a preferred embodiment of the automatic plant identification method of the present invention.
[0070] 在本实施例中, 结合图 1、 图 2和图 3所示, 所述植物自动识别方法运行于智能 设备 1中, 该方法包括如下步骤: [0070] In the embodiment, as shown in FIG. 1, FIG. 2 and FIG. 3, the automatic plant identification method runs on the smart In device 1, the method includes the following steps:
[0071] S10: 获取待识别植物的图像信息; [0071] S10: acquiring image information of the plant to be identified;
[0072] 具体地, 植物图像信息获取模块 100获取待识别植物的图像信息, 所述待识别 植物的图像信息包括待识别植物的叶子、 颈、 花、 果实中的至少两种信息。 其 中叶子的信息至少包括叶片颜色和形状; 颈的信息至少包括颈部颜色和直径参 数; 花的信息至少包括花的形状和颜色; 果实的信息至少包括果实的颜色、 形 状和大小。  Specifically, the plant image information acquiring module 100 acquires image information of the plant to be identified, and the image information of the plant to be identified includes at least two kinds of information of a leaf, a neck, a flower, and a fruit of the plant to be identified. The information of the leaves includes at least the color and shape of the blade; the information of the neck includes at least the color and diameter of the neck; the information of the flower includes at least the shape and color of the flower; and the information of the fruit includes at least the color, shape and size of the fruit.
[0073] S20: 去除待识别植物的图像信息中的噪声;  [0073] S20: removing noise in image information of the plant to be identified;
[0074] 具体地, 图像预处理模块 102去除待识别植物的图像信息中的噪声。 图像预处 理的主要目的是消除图像中无关的信息, 恢复有用的真实信息, 增强有关信息 的可检测性和最大限度地简化数据, 从而改进特征抽取、 图像分割、 匹配和识 别的可靠性。 图像预处理过程一般包括数字化、 归一化、 平滑、 复原和增强等 步骤。  [0074] Specifically, the image pre-processing module 102 removes noise in the image information of the plant to be identified. The main purpose of image preprocessing is to eliminate irrelevant information in the image, restore useful real information, enhance the detectability of the information and minimize the data, thereby improving the reliability of feature extraction, image segmentation, matching and recognition. The image preprocessing process generally includes steps of digitization, normalization, smoothing, restoration, and enhancement.
[0075] 参照图 5所示, 图 5为图 4中步骤 S20的细化流程示意图。 步骤 S20包括如下步骤  Referring to FIG. 5, FIG. 5 is a schematic diagram of the refinement process of step S20 in FIG. Step S20 includes the following steps
[0076] S201 : 将待识别植物的原始图像信息采样并量化得到数字图像。 [0076] S201: Raw image information of the plant to be identified is sampled and quantized to obtain a digital image.
[0077] 具体地, 数字化子模块 1020将待识别植物的原始图像信息采样并量化得到处理 单元 14能够处理的数字图像。 具体地, 一幅待识别植物的原始图像信息 (即原 始图像) 的灰度值是空间变量 (位置的连续值) 的连续函数。 在 MxN点阵上对 该原始图像灰度采样并加以量化 (归为 2b个灰度等级之一) , 可以得到计算机 能够处理的数字图像。 为了使数字图像能重建原来的图像, 对M、 N和 b值的大 小预先设置。 在接收装置的空间和灰度分辨能力范围内, M、 N和 b的数值越大 , 重建图像的质量就越好。 当取样周期等于或小于原始图像中最小细节周期的 一半吋, 重建图像的频谱等于原始图像的频谱, 因此重建图像与原始图像可以 完全相同。 由于 M、 N和 b三者的乘积决定一幅图像在智能终端中的存储量, 因 此在存储量一定的条件下根据图像的不同性质选择合适的M、 N和 b值, 以获取 最好的处理效果。 [0077] Specifically, the digitizing sub-module 1020 samples and quantizes the original image information of the plant to be identified to obtain a digital image that the processing unit 14 can process. Specifically, the gray value of the original image information (i.e., the original image) of a plant to be identified is a continuous function of a spatial variable (continuous value of position). The original image is grayscaled and quantized on the MxN lattice (classified as one of 2b grayscales) to obtain a digital image that the computer can process. In order to enable the digital image to reconstruct the original image, the magnitudes of the M, N, and b values are preset. Within the spatial and grayscale resolution capabilities of the receiving device, the larger the values of M, N, and b, the better the quality of the reconstructed image. When the sampling period is equal to or less than half of the minimum detail period in the original image, the spectrum of the reconstructed image is equal to the spectrum of the original image, so the reconstructed image can be identical to the original image. Since the product of M, N and b determines the storage amount of an image in the intelligent terminal, the appropriate M, N and b values are selected according to the different properties of the image under the condition of a certain amount of storage to obtain the best. Processing effect.
[0078] S202: 通过变换处理保留所述数字图像的特定特征。 [0079] 具体地, 归一化子模块 1022通过变换处理保留所述数字图像的特定特征, 使数 字图像的特定特征在给定变换下具有不变的性质。 图像的特定特征, 例如物体 的面积和周长, 对于坐标旋转变换来说就具有不变的性质。 在一般情况下, 某 些因素或变换对图像一些性质的影响可通过归一化处理得到消除或减弱, 从而 可以被选作测量图像的依据。 例如对于光照不可控的遥感图片, 灰度直方图的 归一化对于图像分析是十分必要的。 灰度归一化、 几何归一化和变换归一化是 获取图像不变性质的三种归一化方法。 [0078] S202: retaining a specific feature of the digital image by a transform process. [0079] Specifically, the normalization sub-module 1022 preserves certain features of the digital image by transform processing such that certain features of the digital image have invariant properties under a given transformation. Certain features of the image, such as the area and perimeter of the object, have invariant properties for coordinate rotation transformations. In general, the influence of certain factors or transformations on some properties of the image can be eliminated or attenuated by normalization, which can be selected as the basis for measuring the image. For example, for remote sensing images with uncontrollable illumination, the normalization of gray histograms is necessary for image analysis. Gray normalization, geometric normalization, and transform normalization are three normalization methods for obtaining the invariant properties of an image.
[0080] S203: 消除所述数字图像中的噪声。  [0080] S203: Eliminate noise in the digital image.
[0081] 具体地, 所述平滑处理子模块 1024消除所述数字图像中的随机噪声。 常用的平 滑方法有中值法、 局部求平均法和 k近邻平均法。 局部区域大小可以是固定的, 也可以是逐点随灰度值大小变化的。 此外, 有吋应用空间频率域带通滤波方法  [0081] Specifically, the smoothing processing sub-module 1024 eliminates random noise in the digital image. Commonly used smoothing methods are median method, local averaging method and k-nearest neighbor averaging method. The size of the local area can be fixed or it can be changed point by point with the size of the gray value. In addition, there is a method for applying spatial frequency domain bandpass filtering
[0082] S204: 校正所述数字图像的像素退化。 [0082] S204: Correcting pixel degradation of the digital image.
[0083] 具体地, 图像复原子模块 1026校正各种原因造成的所述数字图像的像素退化, 使重建或估计得到的图像尽可能逼近于理想无退化的像场。 在实际应用中常常 发生图像像素退化现象。 例如大气流的扰动, 光学系统的像差, 相机和物体的 相对运动都会使遥感图像发生退化。 基本的复原技术是把获取的退化图像 g(x, y )看成是退化函数 h(x, y)和理想图像 f(x, y)的卷积。 它们的傅里叶变换存在关系 G(u, v=H(u, v)F(u, v)。 根据退化机理确定退化函数后, 就可从此关系式求出 F (u, v), 再用傅里叶反变换求出 f(x, y), 通常称为反向滤波器。 当不存在噪声吋 , 维纳滤波器成为理想的反向滤波器。  [0083] Specifically, the image complex atom module 1026 corrects pixel degradation of the digital image caused by various causes, so that the reconstructed or estimated image is as close as possible to the ideal image field without degradation. Image pixel degradation often occurs in practical applications. For example, disturbances in large airflow, aberrations in the optical system, and relative motion of the camera and objects can degrade the remote sensing image. The basic restoration technique is to treat the acquired degraded image g(x, y) as a convolution of the degenerate function h(x, y) and the ideal image f(x, y). Their Fourier transform has a relationship G(u, v=H(u, v)F(u, v). After determining the degradation function according to the degradation mechanism, F (u, v) can be obtained from this relation. Using the inverse Fourier transform to find f(x, y), usually called the inverse filter. When there is no noise, the Wiener filter becomes the ideal inverse filter.
[0084] S205: 对所述数字图像中的信息有选择地进行图像增强处理以改善所述数字图 像的视觉效果。  [0084] S205: Perform image enhancement processing on the information in the digital image to improve the visual effect of the digital image.
[0085] 具体地, 图像增强子模块 1028对所述数字图像中的信息有选择地加强和抑制, 以改善所述数字图像的视觉效果, 或将图像转变为更适合于处理单元 14处理的 形式, 以便于数据抽取或识别。 例如一个图像增强系统可以通过高通滤波器来 突出图像的轮廓线, 从而使机器能够测量轮廓线的形状和周长。 图像增强技术 有多种方法, 反差展宽、 对数变换、 密度分层和直方图均衡等都可用于改变图 像灰调和突出细节。 [0085] Specifically, the image enhancement sub-module 1028 selectively enhances and suppresses information in the digital image to improve the visual effect of the digital image, or to transform the image into a form more suitable for processing by the processing unit 14. For data extraction or identification. For example, an image enhancement system can highlight the outline of an image through a high-pass filter, enabling the machine to measure the shape and perimeter of the outline. There are several ways to enhance image enhancement techniques. Contrast broadening, logarithmic transformation, density stratification, and histogram equalization can all be used to change the graph. Like gray tones and highlighting details.
[0086] S30: 从经过去噪声处理的图像信息中提取植物轮廓特征和颜色特征, 并将提 取的植物轮廓特征和颜色特征分别与预先建立的数据库中的植物轮廓特征和颜 色特征信息进行匹配, 得到待识别植物与数据库中已存在植物信息的匹配度。  [0086] S30: extracting plant outline features and color features from the image information after denoising processing, and matching the extracted plant outline features and color features with the plant outline features and color feature information in a pre-established database, respectively, The degree of matching between the plants to be identified and the existing plant information in the database is obtained.
[0087] 具体地, 参照图 6所示, 图 6为图 4中步骤 S30的细化流程示意图。 步骤 S30包括 如下步骤:  Specifically, referring to FIG. 6, FIG. 6 is a schematic flowchart of the refinement of step S30 in FIG. Step S30 includes the following steps:
[0088] S301 : 采用 Hu矩匹配算法、 轮廓树匹配算法或成对几何直方图匹配算法提取 图像信息中的轮廓特征。  [0088] S301: extracting contour features in the image information by using a Hu moment matching algorithm, a contour tree matching algorithm, or a pairwise geometric histogram matching algorithm.
[0089] 具体地, 轮廓特征提取子模块采用 Hu矩匹配算法、 轮廓树匹配算法或成对几何 直方图匹配算法提取图像信息中的轮廓特征。 轮廓通过轮廓特征的提取和匹配 能够获得植物图像的要素 (例如: 叶子、 颈、 花、 果实) 的形状、 大小等。 由 于轮廓的 Hu矩能够对包括缩放、 旋转和镜像映射在内的变化具有不变性, 因此 在进行轮廓匹配吋通过常用 Hu矩匹配算法。 cvMatchShapes函数可以很方便的实 现对 2个轮廓间的匹配。 轮廓树匹配算法采用树的形式比较两个轮廓。 cvMatchC ontourTrees函数能够实现轮廓树的对比。 成对几何直方图匹配算法是指在得到轮 廓的成对几何直方图之后, 使用直方图对比的方法来进行轮廓匹配。  [0089] Specifically, the contour feature extraction sub-module extracts contour features in the image information by using a Hu moment matching algorithm, a contour tree matching algorithm, or a pairwise geometric histogram matching algorithm. The contour can be extracted and matched by contour features to obtain the shape, size, etc. of the elements of the plant image (for example: leaves, neck, flowers, fruits). Since the Hu moment of the contour can be invariant to changes including scaling, rotation, and mirror mapping, contour matching is performed through a common Hu moment matching algorithm. The cvMatchShapes function makes it easy to match between two contours. The contour tree matching algorithm compares two contours in the form of a tree. The cvMatchC ontourTrees function enables the comparison of contour trees. The pairwise geometric histogram matching algorithm refers to the use of histogram contrast method to perform contour matching after obtaining the paired geometric histogram of the contour.
[0090] S302: 采用颜色矩算法或颜色集算法提取图像信息中的颜色特征。  [0090] S302: Extracting color features in the image information by using a color moment algorithm or a color set algorithm.
[0091] 具体地, 颜色特征提取子模块 1042采用颜色矩算法或颜色集算法提取图像信息 中的颜色特征。 颜色直方图用以反映图像颜色的组成分布, 即各种颜色出现的 概率。 颜色矩算法利用线性代数中矩的概念, 将图像中的颜色分布用颜色矩表 示。 利用颜色一阶矩 (平均值 Average) 、 颜色二阶矩 (方差 Variance) 和颜色 三阶矩 (偏斜度 Skewness) 来描述颜色分布。 与颜色直方图不同, 利用颜色矩进 行图像描述无需量化图像特征。 颜色集算法基于颜色实现对大规模图像的检索 。 该方法将颜色转化到 HSV颜色空间后, 将图像根据其颜色信息进行图像分割 成若干 region, 并将颜色分为多个 bin, 每个 region进行颜色空间量化建立颜色索 弓 I, 进而建立二进制图像颜色索引表。 为加快査找速度, 还可以构造二分査找 树进行特征检索。  [0091] Specifically, the color feature extraction sub-module 1042 extracts color features in the image information using a color moment algorithm or a color set algorithm. The color histogram is used to reflect the composition distribution of the image color, that is, the probability of occurrence of various colors. The color moment algorithm uses the concept of moments in linear algebra to represent the color distribution in an image as a color moment. The color distribution is described by the color first moment (average Average), the color second moment (variance Variance), and the color third moment (skewness). Unlike color histograms, image description using color moments does not require quantization of image features. The color set algorithm implements the retrieval of large-scale images based on color. After the method converts the color into the HSV color space, the image is divided into several regions according to the color information thereof, and the color is divided into multiple bins, and each region performs color space quantization to establish a color cable I, thereby establishing a binary image. Color index table. To speed up the search, you can also construct a binary search tree for feature retrieval.
[0092] S303: 将提取到的图像信息中的轮廓特征和颜色特征后与预先建立的数据库中 的植物轮廓特征和颜色特征信息进行匹配, 得到待识别植物与数据库中已存在 植物信息的匹配度。 [0092] S303: The contour feature and the color feature in the extracted image information are followed by a pre-established database. The plant outline feature and the color feature information are matched to obtain the matching degree of the plant information to be identified and the existing plant information in the database.
[0093] 具体地, 匹配度计算子模块 1044将提取到的图像信息中的轮廓特征和颜色特征 后与预先建立的数据库中的植物轮廓特征和颜色特征信息进行匹配, 得到待识 别植物与数据库中已存在植物信息的匹配度。 具体地, 所述预先建立的数据库 运行于远程服务器中, 智能终端 1通过通讯单元 16与远程服务器建立通讯连接, 从远程服务器的数据库中读取预先存储的植物轮廓特征和颜色特征信息并进行 匹配。  [0093] Specifically, the matching degree calculation sub-module 1044 matches the contour feature and the color feature in the extracted image information with the plant outline feature and the color feature information in the pre-established database to obtain the plant to be identified and the database. The degree of matching of plant information already exists. Specifically, the pre-established database is run in a remote server, and the smart terminal 1 establishes a communication connection with the remote server through the communication unit 16, and reads pre-stored plant outline features and color feature information from the database of the remote server and matches them. .
[0094] S40: 输出匹配度最大的植物信息作为待识别植物的信息。  [0094] S40: Output plant information with the greatest matching degree as information of the plant to be identified.
[0095] 具体地, 植物信息输出模块 106输出匹配度最大的植物信息作为待识别植物的 信息。 该待识别植物的信息至少包括该植物的名称、 图片、 生长习性以及功能 [0095] Specifically, the plant information output module 106 outputs the plant information having the greatest matching degree as the information of the plant to be identified. The information of the plant to be identified includes at least the name, picture, growth habit and function of the plant
, 以供用户参考, 为植物非专业人员在不知道植物名称的情况下提供快速识别 植物信息的途径, 有利于植物非专业人员野外生存。 For the reference of users, it provides a way for plant non-professionals to quickly identify plant information without knowing the name of the plant, which is beneficial to the survival of plant non-professionals in the wild.
[0096] 本发明所述图像匹配和 /或比对过程为现有技术中的图像处理技术, 在此不做 限定和赘述。 The image matching and/or comparison process of the present invention is an image processing technology in the prior art, which is not limited and described herein.
[0097] 本发明提供的植物自动识别方法通过去除待识别植物的图像信息中的噪声, 消 除图像中无关的信息, 恢复有用的真实信息, 通过提取经过预处理的图像信息 的轮廓特征和颜色特征, 并与预先建立的数据库中的植物轮廓特征和颜色特征 信息进行匹配, 得到待识别植物与数据库中已存在植物信息的匹配度, 输出匹 配度最大的植物信息为待识别植物的信息供用户参考, 为植物非专业人员在不 知道植物名称的情况下提供快速识别植物信息的途径, 有利于植物非专业人员 野外生存。  [0097] The automatic plant identification method provided by the invention removes the noise in the image information of the plant to be identified, eliminates irrelevant information in the image, restores useful real information, and extracts contour features and color features of the preprocessed image information. And matching with the plant outline feature and color feature information in the pre-established database, obtaining the matching degree of the existing plant information in the database to be identified, and outputting the plant information with the largest matching degree as the information of the plant to be identified for the user's reference. It provides a way for plant non-professionals to quickly identify plant information without knowing the name of the plant, which is beneficial to the survival of plant non-professionals in the wild.
[0098]  [0098]
[0099] 以上仅为本发明的优选实施例, 并非因此限制本发明的专利范围, 凡是利用本 发明说明书及附图内容所作的等效结构或等效功能变换, 或直接或间接运用在 其他相关的技术领域, 均同理包括在本发明的专利保护范围内。  The above are only the preferred embodiments of the present invention, and are not intended to limit the scope of the invention, and the equivalent structure or equivalent function transformations made by the description of the invention and the drawings are directly or indirectly applied to other related The technical field is equally included in the scope of patent protection of the present invention.
工业实用性  Industrial applicability
[0100] 相较于现有技术, 本发明提供的植物自动识别系统和方法通过去除待识别植物 的图像信息中的噪声, 消除图像中无关的信息, 恢复有用的真实信息, 通过提 取经过预处理的图像信息的轮廓特征和颜色特征, 并与预先建立的数据库中的 植物轮廓特征和颜色特征信息进行匹配, 得到待识别植物与数据库中已存在植 物信息的匹配度, 输出匹配度最大的植物信息为待识别植物的信息供用户参考 , 为植物非专业人员在不知道植物名称的情况下提供快速识别植物信息的途径 , 有利于植物非专业人员野外生存。 [0100] Compared with the prior art, the automatic plant identification system and method provided by the present invention removes the plant to be identified Noise in the image information, eliminating irrelevant information in the image, restoring useful real information, by extracting contour features and color features of the preprocessed image information, and with plant contour features and color feature information in a pre-established database Matching is performed to obtain the matching degree of the plant information in the database to be identified, and the plant information with the largest matching degree is the information of the plant to be identified for the user's reference, and the plant non-professionals provide the fast without knowing the plant name. The way to identify plant information is beneficial to the survival of plant non-professionals in the wild.

Claims

权利要求书 claims
[权利要求 1] 一种植物自动识别系统, 其特征在于, 所述植物自动识别系统运行于 智能终端中, 该系统包括植物图像信息获取模块、 图像预处理模块、 植物特征提取和匹配模块以及植物信息输出模块, 其中: 所述植物图 像信息获取模块用于获取待识别植物的图像信息; 所述图像预处理模 块用于去除待识别植物的图像信息中的噪声; 所述植物特征提取和匹 配模块用于从经过去噪声处理的图像信息中提取植物轮廓特征和颜色 特征, 并将提取的植物轮廓特征和颜色特征分别与预先建立的数据库 中的植物轮廓特征和颜色特征信息进行匹配, 得到待识别植物与数据 库中已存在植物信息的匹配度; 所述植物信息输出模块用于输出匹配 度最大的植物信息作为待识别植物的信息。 [Claim 1] A plant automatic identification system, characterized in that the plant automatic identification system runs in an intelligent terminal, and the system includes a plant image information acquisition module, an image preprocessing module, a plant feature extraction and matching module, and a plant Information output module, wherein: the plant image information acquisition module is used to obtain image information of plants to be identified; the image preprocessing module is used to remove noise in the image information of plants to be identified; the plant feature extraction and matching module Used to extract plant outline features and color features from denoised image information, and match the extracted plant outline features and color features with the plant outline features and color feature information in the pre-established database, respectively, to obtain the plant outline features and color features to be identified The matching degree between the plant and the existing plant information in the database; the plant information output module is used to output the plant information with the largest matching degree as the information of the plant to be identified.
[权利要求 2] 如权利要求 1所述的植物自动识别系统, 其特征在于, 所述图像预处 理模块包括数字化子模块、 归一化子模块、 平滑处理子模块、 图像复 原子模块以及图像增强子模块, 其中: 所述数字化子模块用于将待识 别植物的原始图像信息采样并量化得到数字图像; 所述归一化子模块 用于通过变换处理保留所述数字图像的特定特征; 所述平滑处理子模 块用于消除所述数字图像中的噪声; 图像复原子模块用于校正所述数 字图像的像素退化; 图像增强子模块用于对所述数字图像中的信息进 行图像增强处理以改善所述数字图像的视觉效果。 [Claim 2] The plant automatic identification system according to claim 1, characterized in that the image preprocessing module includes a digitization sub-module, a normalization sub-module, a smoothing sub-module, an image complex atom module and an image enhancement sub-module, wherein: the digitization sub-module is used to sample and quantify the original image information of the plant to be identified to obtain a digital image; the normalization sub-module is used to retain the specific characteristics of the digital image through transformation processing; the The smoothing sub-module is used to eliminate noise in the digital image; the image complex atom module is used to correct the pixel degradation of the digital image; the image enhancement sub-module is used to perform image enhancement processing on the information in the digital image to improve The visual effects of the digital image.
[权利要求 3] 如权利要求 1所述的植物自动识别系统, 其特征在于, 植物特征提取 和匹配模块包括轮廓特征提取子模块、 颜色特征提取子模块以及匹配 度计算子模块, 其中: 所述轮廓特征提取子模块用于采用 Hu矩匹配 算法、 轮廓树匹配算法或成对几何直方图匹配算法提取图像信息中的 轮廓特征; 所述颜色特征提取子模块用于采用颜色矩算法或颜色集算 法提取图像信息中的颜色特征; 所述匹配度计算子模块用于将提取到 的图像信息中的轮廓特征和颜色特征后与预先建立的数据库中的植物 轮廓特征和颜色特征信息进行匹配, 得到待识别植物与数据库中已存 在植物信息的匹配度。 [Claim 3] The plant automatic identification system according to claim 1, wherein the plant feature extraction and matching module includes a contour feature extraction sub-module, a color feature extraction sub-module and a matching degree calculation sub-module, wherein: The contour feature extraction sub-module is used to extract contour features in image information using Hu moment matching algorithm, contour tree matching algorithm or paired geometric histogram matching algorithm; the color feature extraction sub-module is used to use color moment algorithm or color set algorithm Extract the color features in the image information; the matching degree calculation sub-module is used to match the contour features and color features in the extracted image information with the plant contour features and color feature information in the pre-established database to obtain the desired Match the identified plants with existing plant information in the database.
[权利要求 4] 如权利要求 1所述的植物自动识别系统, 其特征在于, 所述待识别植 物的图像信息包括待识别植物的叶子、 颈、 花和果实中的至少两种信 息。 [Claim 4] The automatic plant identification system of claim 1, wherein the image information of the plant to be identified includes at least two types of information of leaves, necks, flowers and fruits of the plant to be identified.
[权利要求 5] 如权利要求 1所述的植物自动识别系统, 其特征在于, 所述待识别植 物的信息至少包括该植物的名称、 图片、 生长习性以及功能。 [Claim 5] The automatic plant identification system of claim 1, wherein the information about the plant to be identified includes at least the name, picture, growth habit and function of the plant.
[权利要求 6] —种植物自动识别方法, 其特征在于, 所述植物自动识别方法运行于 智能终端中, 该方法包括如下步骤: 获取待识别植物的图像信息; 去 除待识别植物的图像信息中的噪声; 从经过去噪声处理的图像信息中 提取植物轮廓特征和颜色特征, 并将提取的植物轮廓特征和颜色特征 分别与预先建立的数据库中的植物轮廓特征和颜色特征信息进行匹配 , 得到待识别植物与数据库中已存在植物信息的匹配度; 输出匹配度 最大的植物信息作为待识别植物的信息。 [Claim 6] A plant automatic identification method, characterized in that the plant automatic identification method is run in an intelligent terminal, and the method includes the following steps: Obtaining the image information of the plant to be identified; Removing the image information of the plant to be identified noise; extract plant outline features and color features from the denoised image information, and match the extracted plant outline features and color features with plant outline features and color feature information in the pre-established database, respectively, to obtain the desired The matching degree between the identified plants and the existing plant information in the database; the plant information with the largest matching degree is output as the information of the plant to be identified.
[权利要求 7] 如权利要求 6所述的植物自动识别方法, 其特征在于, 所述去除待识 别植物的图像信息中的噪声的步骤包括: 将待识别植物的原始图像信 息采样并量化得到数字图像; 通过变换处理保留所述数字图像的特定 特征; 消除所述数字图像中的噪声; 校正所述数字图像的像素退化; 对所述数字图像中的信息进行图像增强处理以改善所述数字图像的视 觉效果。 [Claim 7] The automatic identification method of plants according to claim 6, wherein the step of removing noise from the image information of the plant to be identified includes: sampling and quantifying the original image information of the plant to be identified to obtain a digital Image; retain specific characteristics of the digital image through transformation processing; eliminate noise in the digital image; correct pixel degradation of the digital image; perform image enhancement processing on the information in the digital image to improve the digital image visual effects.
[权利要求 8] 如权利要求 6所述的植物自动识别方法, 其特征在于, 所述从经过去 噪声处理的图像信息中提取植物轮廓特征和颜色特征, 并将提取的植 物轮廓特征和颜色特征分别与预先建立的数据库中的植物轮廓特征和 颜色特征信息进行匹配, 得到待识别植物与数据库中已存在植物信息 的匹配度的步骤包括: 采用 Hu矩匹配算法、 轮廓树匹配算法或成对 几何直方图匹配算法提取图像信息中的轮廓特征; 采用颜色矩算法或 颜色集算法提取图像信息中的颜色特征; 将提取到的图像信息中的轮 廓特征和颜色特征后与预先建立的数据库中的植物轮廓特征和颜色特 征信息进行匹配, 得到待识别植物与数据库中已存在植物信息的匹配 度。 [Claim 8] The automatic plant identification method according to claim 6, characterized in that: extracting plant outline features and color features from the image information that has undergone denoising processing, and using the extracted plant outline features and color features Match the plant outline features and color feature information in the pre-established database respectively. The steps to obtain the matching degree of the plant to be identified and the existing plant information in the database include: using Hu moment matching algorithm, contour tree matching algorithm or pairwise geometry The histogram matching algorithm extracts the contour features in the image information; the color moment algorithm or the color set algorithm is used to extract the color features in the image information; the extracted contour features and color features in the image information are compared with the plants in the pre-established database The contour features and color feature information are matched to obtain the matching degree between the plant to be identified and the existing plant information in the database.
[权利要求 9] 如权利要求 6所述的植物自动识别方法, 其特征在于, 所述待识别植 物的图像信息包括待识别植物的叶子、 颈、 花和果实中的至少两种信 息。 [Claim 9] The automatic plant identification method of claim 6, wherein the image information of the plant to be identified includes at least two types of information of leaves, necks, flowers and fruits of the plant to be identified.
[权利要求 10] 如权利要求 6所述的植物自动识别方法, 其特征在于, 所述待识别植 物的信息至少包括该植物的名称、 图片、 生长习性以及功能。 [Claim 10] The automatic plant identification method according to claim 6, wherein the information about the plant to be identified at least includes the name, picture, growth habit and function of the plant.
PCT/CN2016/108821 2016-06-17 2016-12-07 Automatic plant identification system and method WO2017215206A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201610437411.6 2016-06-17
CN201610437411.6A CN106096563A (en) 2016-06-17 2016-06-17 Plant automatic recognition system and method

Publications (1)

Publication Number Publication Date
WO2017215206A1 true WO2017215206A1 (en) 2017-12-21

Family

ID=57236470

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/108821 WO2017215206A1 (en) 2016-06-17 2016-12-07 Automatic plant identification system and method

Country Status (2)

Country Link
CN (1) CN106096563A (en)
WO (1) WO2017215206A1 (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109558883A (en) * 2018-12-03 2019-04-02 宁夏智启连山科技有限公司 Leaf characteristic extracting method and device
CN112906656A (en) * 2021-03-30 2021-06-04 自然资源部第三海洋研究所 Underwater photo coral reef recognition method, system and storage medium
CN113111672A (en) * 2021-04-13 2021-07-13 中国科学院东北地理与农业生态研究所 Method for judging true wetland plants
CN113188662A (en) * 2021-03-16 2021-07-30 云南电网有限责任公司玉溪供电局 Infrared thermal imaging fault automatic identification system and method
CN117611885A (en) * 2023-11-17 2024-02-27 贵州省生物研究所 Waiting bird ecological regulation and control method based on Canny edge detection

Families Citing this family (17)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106096563A (en) * 2016-06-17 2016-11-09 深圳市易特科信息技术有限公司 Plant automatic recognition system and method
CN106776877B (en) * 2016-11-28 2020-05-01 常州市星网计算机技术有限公司 Automatic identification method for motorcycle part models
CN106503695B (en) * 2016-12-02 2019-07-09 汕头大学 A kind of tobacco plant identification and method of counting based on Aerial Images
CN106596863A (en) * 2016-12-22 2017-04-26 邬东强 System and method for acquiring plant information based on camera of mobile phone
CN106614484A (en) * 2017-01-19 2017-05-10 成都市宏德永兴养殖有限公司 Vehicle-mounted fruit tree pesticide spraying method
CN106934416B (en) * 2017-02-23 2021-03-30 广州讯动网络科技有限公司 Big data-based model matching method
CN107633038A (en) * 2017-09-13 2018-01-26 浙江大学 Tealeaves recognition methods and its system based on image recognition technology
CN109729236A (en) * 2017-10-27 2019-05-07 富泰华工业(深圳)有限公司 Diving recording device
CN108564583B (en) * 2018-04-23 2020-12-01 广东云林信息工程技术股份有限公司 Sample plot monitoring method, device and system
CN109409449A (en) * 2018-09-10 2019-03-01 天津大学 A kind of mobile terminal that can identify Chinese medicinal material
CN110278630B (en) * 2019-05-22 2022-09-27 杭州极致光生物照明有限公司 Spectrum color-modulation illumination method and device and aquarium lamp thereof
CN111178437A (en) * 2019-12-30 2020-05-19 深圳市芭田生态工程股份有限公司 Fertilizing method and computer readable storage medium
CN113191175A (en) * 2020-01-14 2021-07-30 靳爱丛 Information reminding platform and method
CN113407755A (en) * 2020-03-17 2021-09-17 北京百度网讯科技有限公司 Plant growth condition information acquisition method and device and electronic equipment
CN112580493B (en) * 2020-12-16 2021-11-09 广东省林业科学研究院 Plant identification method, device and equipment based on unmanned aerial vehicle remote sensing and storage medium
CN112579804A (en) * 2020-12-28 2021-03-30 西安航空学院 Toxic plant detector
CN113808139A (en) * 2021-09-15 2021-12-17 南京思辨力电子科技有限公司 Intelligent image identification method for Internet of things

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102521600A (en) * 2011-11-03 2012-06-27 北京农业信息技术研究中心 Method and system for identifying white-leg shrimp disease on basis of machine vision
US9135520B2 (en) * 2009-05-19 2015-09-15 Digimarc Corporation Histogram methods and systems for object recognition
CN105203456A (en) * 2015-10-28 2015-12-30 小米科技有限责任公司 Plant species identification method and apparatus thereof
CN106096563A (en) * 2016-06-17 2016-11-09 深圳市易特科信息技术有限公司 Plant automatic recognition system and method

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9135520B2 (en) * 2009-05-19 2015-09-15 Digimarc Corporation Histogram methods and systems for object recognition
CN102521600A (en) * 2011-11-03 2012-06-27 北京农业信息技术研究中心 Method and system for identifying white-leg shrimp disease on basis of machine vision
CN105203456A (en) * 2015-10-28 2015-12-30 小米科技有限责任公司 Plant species identification method and apparatus thereof
CN106096563A (en) * 2016-06-17 2016-11-09 深圳市易特科信息技术有限公司 Plant automatic recognition system and method

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
ZHOU, FALV ET AL.: "Design and Implementation of Online Image Recognition System of Chinese Herbal Medicine Leaf", COMPUTER KNOWLEDGE AND TECHNOLOGY, vol. 10, no. 13, 31 May 2014 (2014-05-31), pages 3114, ISSN: 1009-3044 *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109558883A (en) * 2018-12-03 2019-04-02 宁夏智启连山科技有限公司 Leaf characteristic extracting method and device
CN109558883B (en) * 2018-12-03 2023-04-18 宁夏智启连山科技有限公司 Blade feature extraction method and device
CN113188662A (en) * 2021-03-16 2021-07-30 云南电网有限责任公司玉溪供电局 Infrared thermal imaging fault automatic identification system and method
CN112906656A (en) * 2021-03-30 2021-06-04 自然资源部第三海洋研究所 Underwater photo coral reef recognition method, system and storage medium
CN113111672A (en) * 2021-04-13 2021-07-13 中国科学院东北地理与农业生态研究所 Method for judging true wetland plants
CN117611885A (en) * 2023-11-17 2024-02-27 贵州省生物研究所 Waiting bird ecological regulation and control method based on Canny edge detection

Also Published As

Publication number Publication date
CN106096563A (en) 2016-11-09

Similar Documents

Publication Publication Date Title
WO2017215206A1 (en) Automatic plant identification system and method
Parisi-Baradad et al. Otolith shape contour analysis using affine transformation invariant wavelet transforms and curvature scale space representation
Li et al. Immature green citrus fruit detection and counting based on fast normalized cross correlation (FNCC) using natural outdoor colour images
CN103822879B (en) A kind of Fructus actinidiae chinensis based on high light spectrum image-forming technology expand fruit lossless detection method
Hati et al. Plant recognition from leaf image through artificial neural network
CN110188614A (en) It is a kind of based on skin crack segmentation NLM filtering refer to vein denoising method
CN108564092A (en) Sunflower disease recognition method based on SIFT feature extraction algorithm
CN110008912B (en) Social platform matching method and system based on plant identification
Tripathy Detection of cotton leaf disease using image processing techniques
Song et al. Shadow removal method of apples based on fuzzy set theory
Kim et al. Tree recognition for landscape using by combination of features of its leaf, flower and bark
Wang et al. A single scale retinex based method for palm vein extraction
Shi et al. Group-based sparse representation for low lighting image enhancement
CN116468958B (en) Communication tower safety detection method and system
Wang et al. Application of support vector machine to apple recognition using in apple harvesting robot
US10115195B2 (en) Method and apparatus for processing block to be processed of urine sediment image
CN114758139B (en) Method for detecting accumulated water in foundation pit
Mu et al. Tree leaf feature extraction and recognition based on geometric features and haar wavelet theory
Khan et al. Shadow removal from digital images using multi-channel binarization and shadow matting
CN110019904A (en) A kind of image search method based on big data
Sridevy et al. Image pre-processing techniques utilized for the plant identification: A review
Tamilselvi et al. Color based K-Means Clustering For Image Segmentation to Identify the Infected Leaves
Wei et al. Green apple recognition in natural illumination based on random forest algorithm
CN109034129A (en) A kind of robot with face identification functions
Deepa A Pre Processing Approach for Accurate Identification of Plant Diseases in leaves

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 16905318

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 16905318

Country of ref document: EP

Kind code of ref document: A1