CN116385375A - Forest defect area detection method and device based on remote sensing image and storage medium - Google Patents
Forest defect area detection method and device based on remote sensing image and storage medium Download PDFInfo
- Publication number
- CN116385375A CN116385375A CN202310261889.8A CN202310261889A CN116385375A CN 116385375 A CN116385375 A CN 116385375A CN 202310261889 A CN202310261889 A CN 202310261889A CN 116385375 A CN116385375 A CN 116385375A
- Authority
- CN
- China
- Prior art keywords
- image
- characteristic information
- forest
- image block
- remote sensing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 230000007547 defect Effects 0.000 title claims abstract description 72
- 238000001514 detection method Methods 0.000 title claims abstract description 19
- 238000000034 method Methods 0.000 claims description 55
- 230000002950 deficient Effects 0.000 claims description 51
- 230000008569 process Effects 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 6
- 230000000694 effects Effects 0.000 abstract description 10
- 238000010586 diagram Methods 0.000 description 20
- 239000011159 matrix material Substances 0.000 description 15
- 239000013598 vector Substances 0.000 description 15
- 238000013527 convolutional neural network Methods 0.000 description 12
- 230000010354 integration Effects 0.000 description 9
- 230000006870 function Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 230000000593 degrading effect Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10032—Satellite or aerial image; Remote sensing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30181—Earth observation
- G06T2207/30188—Vegetation; Agriculture
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Quality & Reliability (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The application discloses a forest defect area detection method, device and storage medium based on remote sensing images, comprising the following steps: determining the position information of the target forest, and determining a remote sensing image corresponding to the target forest according to the position information of the target forest; in the remote sensing image, determining a forest image area corresponding to a target forest; dividing a forest image area into a plurality of image blocks with the same size; generating first characteristic information corresponding to each image block according to the pixel value of each image block; generating second characteristic information corresponding to the first characteristic information according to the correlation between the first characteristic information corresponding to each image block; and judging whether the target forest has defects according to the second characteristic information. Thereby achieving the technical effect of accurately identifying the defect area in the forest.
Description
Technical Field
The application relates to the field of satellite monitoring, in particular to a forest defect area detection method and device based on remote sensing images and a storage medium.
Background
At present, environmental problems are continuously attracting attention from the country, and the country has put more effort on forest construction. Besides the forest area is expanded by returning and cultivating the forest, a plurality of places continue to plant the forest on the basis of the original forest, thereby expanding the forest area. Many forests are thus formed on the basis of multiple planned planting.
But in the forest planting process, the planting process often needs to be supervised. There are the following problems in terms of supervision: in order to cheat supervision, trees inconsistent with planning can be planted in the forest expansion planting process to cheat supervision, or incorrect trees can be planted by staff in the forest expansion planting process due to lack of expertise. Over time, however, the growth of these trees tends to be undesirable, thereby creating defective areas in the forest.
Although a method for analyzing whether a forest has a defective area by using a forest remote sensing image is proposed in the prior art, the existing analysis method generally inputs the remote sensing image of the whole forest into a trained convolutional neural network, so as to identify the defective area in the forest. The convolutional neural network usually convolves the whole remote sensing image by convolution check to generate a Feature Map (Feature Map), and performs Feature extraction by using the Feature Map. Therefore, the features extracted by using the convolutional neural network are still extracted based on the texture, color and other information of the remote sensing image. The feature map thus reflects the overall features of the remote sensing image, which do not reflect the associated features between the various image areas in the remote sensing image.
Whereas defective areas in forests often appear to grow poorly, rather than completely degrading to sandy or bare land. It is therefore difficult to accurately identify defective areas in a forest unless the difference features between defective areas and normal areas are deeply mined and compared. Because the existing method for inputting the remote sensing image into the convolutional neural network to identify the defect region in the forest is difficult to mine the difference characteristics between the defect region and the normal region, the defect region in the forest is difficult to accurately identify.
The publication number is CN115512231A, and the name is a remote sensing interpretation method suitable for ecological restoration of the homeland space. Comprising the following steps: acquiring a gray scale interval; obtaining the contrast ratio of the pixel corresponding to each gray level and the pixel in the 8 neighborhood of the pixel according to the number of the pixel corresponding to each gray level and the gray value of each pixel in the 8 neighborhood of the pixel corresponding to each gray level; obtaining contrast in a gray scale interval; acquiring an expected enhancement effect of gray level in a gray level interval; obtaining an optimal division mode according to the expected enhancement effect of the gray level in each gray level interval after each division, carrying out histogram equalization on the gray level histogram of the remote sensing image of the ecological restoration area according to the optimal division mode to obtain an optimal enhancement image, and judging the ecological restoration effect of the homeland space according to the optimal enhancement image.
The publication number is CN114332346A, and the name is a method for carrying out forest ecological function division by utilizing a remote sensing technology. The method specifically comprises the following steps: step one, remote sensing image processing; step two, stability analysis; step three, standard drafting; step four, planning a forest; step five, ecological implementation; the invention relates to the technical field of ecological functional division.
Aiming at the technical problems that in the prior art, as the existing mode of inputting the remote sensing image into the convolutional neural network so as to identify the defect area in the forest, the difference characteristics between the defect area and the normal area are difficult to excavate, and therefore, the defect area in the forest is difficult to accurately identify, no effective solution is proposed at present.
Disclosure of Invention
The embodiment of the disclosure provides a forest defect area detection method, device and storage medium based on a remote sensing image, which at least solve the technical problems that in the prior art, as the existing mode of inputting the remote sensing image into a convolutional neural network so as to identify a defect area in a forest, the difference characteristics between the defect area and a normal area are difficult to excavate, and therefore the defect area in the forest is difficult to accurately identify.
According to an aspect of the embodiments of the present disclosure, there is provided a forest defect area detection method based on a remote sensing image, including: determining the position information of the target forest, and determining a remote sensing image corresponding to the target forest according to the position information of the target forest; in the remote sensing image, determining a forest image area corresponding to a target forest; dividing a forest image area into a plurality of image blocks with the same size, wherein each image block contains a corresponding pixel value; generating first characteristic information corresponding to each image block according to the pixel value of each image block, wherein the first characteristic information is used for indicating the image characteristics of the corresponding image block; generating second characteristic information corresponding to the first characteristic information according to the correlation between the first characteristic information corresponding to each image block, wherein the second characteristic information is used for indicating the image characteristics of all the image blocks; and judging whether the target forest has defects according to the second characteristic information.
According to another aspect of the embodiments of the present disclosure, there is also provided a storage medium including a stored program, wherein the method of any one of the above is performed by a processor when the program is run.
According to another aspect of the embodiments of the present disclosure, there is also provided a forest defect area detection device based on a remote sensing image, including: the position information determining module is used for determining the position information of the target forest and determining a remote sensing image corresponding to the target forest according to the position information of the target forest; the forest image area determining module is used for determining a forest image area corresponding to the target forest in the remote sensing image; the image block dividing module is used for dividing the forest image area into a plurality of image blocks with the same size, wherein each image block contains a corresponding pixel value; the first characteristic information generation module is used for generating first characteristic information corresponding to each image block according to the pixel value of each image block, wherein the first characteristic information is used for indicating the image characteristics of the corresponding image block; the second characteristic information generation module is used for generating second characteristic information corresponding to the first characteristic information respectively according to the correlation among the first characteristic information corresponding to each image block, wherein the second characteristic information is used for indicating the image characteristics of all the image blocks; and the judging module is used for judging whether the target forest has defects according to the second characteristic information.
According to another aspect of the embodiments of the present disclosure, there is also provided a forest defect area detection device based on a remote sensing image, including: a processor; and a memory, coupled to the processor, for providing instructions to the processor for processing the steps of: determining the position information of the target forest, and determining a remote sensing image corresponding to the target forest according to the position information of the target forest; in the remote sensing image, determining a forest image area corresponding to a target forest; dividing a forest image area into a plurality of image blocks with the same size, wherein each image block contains a corresponding pixel value; generating first characteristic information corresponding to each image block according to the pixel value of each image block, wherein the first characteristic information is used for indicating the image characteristics of the corresponding image block; generating second characteristic information corresponding to the first characteristic information according to the correlation between the first characteristic information corresponding to each image block, wherein the second characteristic information is used for indicating the image characteristics of all the image blocks; and judging whether the target forest has defects according to the second characteristic information.
Firstly, the remote sensing data platform determines the position information of a target forest, and determines a remote sensing image corresponding to the target forest according to the position information of the target forest. And then, the remote sensing data platform determines a forest image area corresponding to the target forest in the remote sensing image. Further, the remote sensing data platform divides the forest image area into a plurality of image blocks with the same size, and generates first characteristic information corresponding to each image block according to pixel values of each image block. In addition, the remote sensing data platform generates second characteristic information corresponding to the first characteristic information according to the correlation between the first characteristic information corresponding to each image block. And finally, the remote sensing data platform judges whether the target forest has defects according to the second characteristic information.
Since the specified image block in the present application has a correlation with other image blocks than the specified image block, the second feature information includes not only the image features of the specified image block (for example, the image features of the image block corresponding to the defective area) but also the image features of the other image blocks than the specified image block (for example, the image features of the image block corresponding to the normal area). In addition, since the present application determines whether the target forest has a defect according to the second feature information, the difference between the image block of the defective area and the image block of the normal area can be better represented by comparing the second feature information of the designated image block (for example, the image feature of the image block corresponding to the defective area) with the second feature information of the other image blocks except for the designated image block (for example, the image feature of the image block corresponding to the normal area). Therefore, the technical effect of accurately identifying the defect area in the forest is achieved through the operation. The method solves the technical problems that in the prior art, as the existing method of inputting the remote sensing image into the convolutional neural network so as to identify the defective area in the forest, the difference characteristics between the defective area and the normal area are difficult to excavate, and the defective area in the forest is difficult to accurately identify.
Drawings
The accompanying drawings, which are included to provide a further understanding of the disclosure and are incorporated in and constitute a part of this application, illustrate embodiments of the disclosure and together with the description serve to explain the disclosure and do not constitute an undue limitation on the disclosure. In the drawings:
FIG. 1 is a schematic diagram of a remote control and detection system according to a first aspect of embodiment 1 of the present application;
FIG. 2 is a flow chart of a method for remote sensing image based forest defect area detection according to a first aspect of embodiment 1 of the present application;
fig. 3 is a schematic diagram of a remote sensing image of a forest image area corresponding to a target forest according to the first aspect of embodiment 1 of the present application;
FIG. 4 is a block diagram of a first aspect of embodiment 1 of the present application 1 ~Z m Is a schematic diagram of a forest image area;
FIG. 5A is a first image block Z according to a first aspect of embodiment 1 of the present application 1 First color channel Zr of (1) 1 Is a pixel schematic of (1);
FIG. 5B is a second image block Z according to the first aspect of embodiment 1 of the present application 2 First color channel Zr of (1) 2 Is a pixel schematic of (1);
FIG. 5C is an mth image block Z according to the first aspect of embodiment 1 of the present application m First color channel Zr of (1) m Is a pixel schematic of (1);
FIG. 6 is a diagram of a remote sensing data platform according to a first aspect of embodiment 1 of the present application for providing second characteristic information B 1 ~B m Splicing the two images into a schematic diagram of a multidimensional matrix;
FIG. 7 is a schematic diagram of a defect region identification model according to a first aspect of embodiment 1 of the present application;
FIG. 8 is a schematic diagram of a convolution kernel of a first convolution layer according to a first aspect of embodiment 1 of the present application;
FIG. 9 is a schematic illustration of a fully connected layer and a softmax sorting layer according to the first aspect of example 1 of the present application;
FIG. 10 is a schematic block diagram of a forest defect area detection device based on remote sensing images according to a first aspect of embodiment 2 of the present application; and
fig. 11 is a schematic block diagram of a forest defect area detection device based on a remote sensing image according to the first aspect of embodiment 3 of the present application.
Detailed Description
In order to better understand the technical solutions of the present disclosure, the following description will clearly and completely describe the technical solutions of the embodiments of the present disclosure with reference to the drawings in the embodiments of the present disclosure. It will be apparent that the described embodiments are merely embodiments of a portion, but not all, of the present disclosure. All other embodiments, which can be made by one of ordinary skill in the art without inventive effort, based on the embodiments in this disclosure, shall fall within the scope of the present disclosure.
It should be noted that the terms "first," "second," and the like in the description and claims of the present disclosure and in the foregoing figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the disclosure described herein may be capable of operation in sequences other than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
Example 1
In accordance with the present embodiment, it is provided that the steps illustrated in the flow diagrams of the figures may be performed in a computer system, such as a set of computer-executable instructions, and that although a logical order is illustrated in the flow diagrams, in some cases the steps illustrated or described may be performed in an order other than that illustrated herein.
Fig. 1 is a schematic diagram of a remote control and detection system according to an embodiment of the present application. Referring to fig. 1, the system includes: remote sensing data platform 10, remote sensing image transmission system 20 and terminal equipment 30. Wherein the remote sensing image transmission system 20 includes a satellite 21 and a drone 22. And wherein the satellite 21 and the unmanned aerial vehicle 22 are provided with an image acquisition module configured to acquire remote sensing images of the ground. Specifically, the satellite 21 may be an ultra low orbit satellite, so that a high definition remote sensing image may be photographed. Then, the satellite 21 and the unmanned aerial vehicle 22 transmit the photographed remote sensing image to the remote sensing data platform 10, and a processor is provided in the remote sensing data platform 10 so as to be processed by the remote sensing data platform 10. Furthermore, the terminal device 30 may interact with the telemetry data platform 10, for example, via a network.
In the above-described operating environment, according to a first aspect of the present embodiment, there is provided a method for detecting a forest defect area based on a remote sensing image, which is implemented, for example, by the remote sensing data platform 10 in fig. 1. Fig. 2 shows a schematic flow chart of the method, and referring to fig. 2, the method includes:
s202: determining the position information of the target forest, and determining a remote sensing image corresponding to the target forest according to the position information of the target forest;
S204: in the remote sensing image, determining a forest image area corresponding to a target forest;
s206: dividing a forest image area into a plurality of image blocks with the same size, wherein each image block contains a corresponding pixel value;
s208: generating first characteristic information corresponding to each image block according to the pixel value of each image block, wherein the first characteristic information is used for representing the image characteristics of the appointed image block;
s210: generating second characteristic information corresponding to the first characteristic information according to the correlation between the first characteristic information corresponding to each image block, wherein the second characteristic information is used for indicating the image characteristics of all the image blocks; and
s212: and judging whether the target forest has defects according to the second characteristic information.
First, the remote sensing data platform 10 determines the position information of the target forest, and determines a remote sensing image corresponding to the target forest based on the position information of the target forest (S202).
Specifically, to implement supervision of the target forest, the staff may send, for example, land planning information related to the target forest to the remote sensing data platform 10 via the terminal device 30. The remote sensing data platform 10 thus obtains land planning information associated with the target forest. The land planning information may be, for example, a land planning scheme corresponding to the target forest, which is formulated by the relevant planning department. And, the land planning scheme includes location information of the target forest.
Thus, the remote sensing data platform 10 may determine the location information of the target forest according to a pre-established land planning scheme.
Then, after acquiring the land planning information, the remote sensing data platform 10 determines a remote sensing image corresponding to the target forest to be supervised according to the position information included in the land planning information. The remote sensing image may be, for example, a remote sensing image comprising a plurality of color channels, each color channel corresponding to a different spectral band. For example, the remote sensing image includes three different color channels of RGB.
Then, the remote sensing data platform 10 determines a forest image area corresponding to the target forest in the remote sensing image (S204). Specifically, since in the remote sensing image, in addition to the forest image area corresponding to the target forest, for example, a forest image area corresponding to another forest is included. The remote sensing data platform 10 thus determines a forest image area corresponding to the target forest on the remote sensing image based on the acquired land planning information after determining the remote sensing image.
Fig. 3 is a schematic diagram of a remote sensing image of a forest image area corresponding to a target forest according to an embodiment of the present application. Referring to fig. 3, the remote sensing image 100 includes a forest image area 110 corresponding to a target forest. And further, the remote sensing image 100 may be, for example, a remote sensing image including three color channels of RGB.
Further, the remote sensing data platform 10 divides the forest image area corresponding to the target forest into a plurality of image blocks Z with the same size 1 ~Z m (S206). Specifically, FIG. 4 is a block Z divided into a plurality of equally sized image blocks according to an embodiment of the present application 1 ~Z m Is a schematic representation of a forest image area 110. Referring to fig. 4, the forest image area 110 is divided into a plurality of image blocks Z of the same size 1 ~Z m . And although not shown in fig. 4, each image block Z 1 ~Z m Each comprising 3 color channels, respectively a first color channel Zr 1 ~Zr m Second color channel Zg 1 ~Zg m Third color channel Zb 1 ~Zb m 。
Furthermore, each image block Z 1 ~Z m Each color channel of (1) includes n pixels. FIG. 5A is a first image block Z according to an embodiment of the present application 1 First color channel Zr of (1) 1 Fig. 5B is a schematic diagram of a second image block Z according to an embodiment of the present application 2 First color channel Zr of (1) 2 And FIG. 5C is an illustration of an mth image block Z according to an embodiment of the present application m First color channel Zr of (1) m Is a pixel schematic of (c). Referring to fig. 5A, a first image block Z 1 First color channel Zr of (1) 1 Comprises n pixels, wherein Zr 1,1 ~Zr 1,n Pixel values corresponding to different pixels, respectively. Referring to FIG. 5B, a second image block Z 2 First color channel Zr of (1) 2 Comprises n pixels, wherein Zr 2,1 ~Zr 2,n Pixel values corresponding to different pixels, respectively. Referring to FIG. 5C, the mth image block Z m First color channel Zr of (1) m Comprises n pixels, wherein Zr m,1 ~Zr m,n Pixel values corresponding to different pixels, respectively.
Similarly, a first image block Z 1 Second color channel Zg of (2) 1 Comprising n pixels, where Zg 1,1 ~Zg 1,n Pixel values corresponding to different pixels, respectively. Second image block Z 2 Second color channel Zg of (2) 2 Also comprises n pixels, wherein Zg 2,1 ~Zg 2,n Pixel values corresponding to different pixels respectively, and so on, mth image block Z m Second color channel Zg of (2) m Also comprises n pixels, wherein Zg m,1 ~Zg m,n Pixel values corresponding to different pixels, respectively.
First image block Z 1 Third color channel Zb of (2) 1 Comprising n pixels, where Zb 1,1 ~Zb 1,n Pixel values corresponding to different pixels, respectively. Second pixel block Z 2 Third color channel Zb of (2) 2 Also contains n pixels, wherein Zb 2,1 ~Zb 2,n Pixel values corresponding to different pixels respectively, and so onPush, mth pixel block Z m Third color channel Zb of (2) m Also contains n pixels, wherein Zb m,1 ~Zb m,n Pixel values corresponding to different pixels, respectively.
Furthermore, the remote sensing data platform 10 is based on the respective image blocks Z 1 ~Z m Generates and generates pixel values corresponding to the respective image blocks Z 1 ~Z m Corresponding first characteristic information A 1 ~A m (S208). Specifically, for image block Z 1 ~Z m Each image block in the plurality of image blocks is connected with each row in each color channel in sequence from beginning to end to form first characteristic information A corresponding to each image block 1 ~A m . Furthermore, the first characteristic information A of each image block 1 ~A m Respectively, feature vectors corresponding to different color channels. For example, a first image block Z 1 First characteristic information a of (a) 1 The method comprises the following steps of: ar (Ar) 1 、Ag 1 Ab 1 Second image block Z 2 First characteristic information a of (a) 2 The method comprises the following steps of: ar (Ar) 2 、Ag 2 Ab 2 Third image block Z 3 First characteristic information a of (a) 3 The method comprises the following steps of: ar (Ar) 3 、Ag 3 Ab 3 Similarly, the mth image block Z m First characteristic information a of (a) m The method comprises the following steps of: ar (Ar) m 、Ag m Ab m 。
For example, first characteristic information A 1 The feature vectors of (2) may be represented as follows: ar (Ar) 1 =[ar 1,1 , ar 1,2 , ar 1,3 ,..., ar 1,n ] T =[Zr 1,1 , Zr 1,2 , Zr 1,3 ,..., Zr 1,n ] T ，Ag 1 =[ag 1,1 , ag 1,2 , ag 1,3 ,..., ag 1,n ] T =[Zg 1,1 , Zg 1,2 , Zg 1,3 ,..., Zg 1,n ] T Ab 1 =[ab 1,1 , ab 1,2 , ab 1,3 ,..., ab 1,n ] T =[Zb 1,1 , Zb 1,2 , Zb 1,3 ,..., Zb 1,n ] T 。
By analogy, the remote sensing data platform 10 generates an image block Z 2 ~Z m First characteristic information a of (a) 2 ~A m . And will not be described in detail herein.
The remote sensing data platform 10 then generates a plurality of images based on the image block Z 1 ~Z m Corresponding first characteristic information A 1 ~A m Correlation with the first characteristic information A 1 ~A m Corresponding second characteristic information B 1 ~B m (S210). Specifically, with the first characteristic information A 1 ~A m Similarly, each second characteristic information B 1 ~B m And also feature vectors corresponding to the respective color channels, respectively. For example with the first image block Z 1 Corresponding second characteristic information B 1 Comprising the following steps: br (Br) 1 、Bg 1 Bb 1 And a second image block Z 2 Corresponding second characteristic information B 2 Comprising the following steps: br (Br) 2 、Bg 2 Bb 2 And a third image block Z 3 Corresponding second characteristic information B 3 Comprising the following steps: br (Br) 3 、Bg 3 Bb 3 And the like, with the mth image block Z m Corresponding second characteristic information B m Comprising the following steps: br (Br) m 、Bg m Bb m 。
Further, the remote sensing data platform 10 is based on the first characteristic information a 1 ~A m Correlation generation between the first characteristic information A and the second characteristic information A 1 ~A m Corresponding second characteristic information B 1 ~B m The detailed steps of (a) will be described in detail later, and will not be described in detail here.
Finally, the remote sensing data platform 10 generates a second characteristic information B 1 ~B m It is determined whether or not the target forest is defective (S212).
As described in the background, defective areas in forests often appear to grow poorly, rather than completely degrading to sandy or bare land. It is therefore difficult to accurately identify defective areas in a forest unless the difference features between defective areas and normal areas are deeply mined and compared. Because the existing method for inputting the remote sensing image into the convolutional neural network to identify the defect region in the forest is difficult to mine the difference characteristics between the defect region and the normal region, the defect region in the forest is difficult to accurately identify.
In view of the above, the application discloses a forest defect area detection method based on remote sensing images. Whereas, since the specified image block in the present application has a correlation with other image blocks than the specified image block, the second characteristic information B 1 ~B m Not only the image features of the specified image block (for example, the image features of the image block corresponding to the defective area) but also the image features of the image blocks other than the specified image block (for example, the image features of the image block corresponding to the normal area) are included. And because the application is based on the second characteristic information B 1 ~B m Determining whether the target forest is defective, thereby comparing the second characteristic information B of the designated image block 1 ~B m (may be, for example, the image feature of the image block corresponding to the defective region), and the second feature information B of the other image blocks than the specified image block 1 ~B m (for example, the image features of the image blocks corresponding to the normal region) can better represent the difference between the image blocks of the defective region and the image blocks of the normal region. Therefore, the technical effect of accurately identifying the defect area in the forest is achieved through the operation. And further solves the technical problem that the difference characteristics between the defective area and the normal area are difficult to excavate because the conventional method of inputting the remote sensing image 100 into the convolutional neural network so as to identify the defective area in the forest in the prior art, so that the defective area in the forest is difficult to accurately identify.
Optionally, the operation of determining whether the target forest has a defect according to the second feature information includes: determining a defect area identification model; and inputting the second characteristic information into the defect area identification model, and judging whether the target forest has defects or not.
Specifically, first, the remote sensing data platform 10 will send the second characteristic information B 1 ~B m Is input to a predetermined defective area identification model 200. While the second characteristic information B is displayed on the remote sensing data platform 10 1 ~B m Before being input into the predetermined defective area identification model 200, the second characteristic information B is required 1 ~B m Spliced into a multi-dimensional matrix (i.e., a feature matrix). FIG. 6 is a diagram of the remote sensing data platform 10 according to an embodiment of the present application for providing second characteristic information B 1 ~B m And splicing the two images into a schematic diagram of the multidimensional matrix. As shown with reference to FIG. 6, due to each of the second characteristic information B 1 ~B m Also respectively comprises the characteristic vectors corresponding to the color channels, so the remote sensing data platform 10 will send the second characteristic information B 1 ~B m And the feature vectors corresponding to the color channels are spliced to form a multidimensional matrix. For example, the second characteristic information B 1 ~B m The matrix corresponding to the first color channel is: br (Br) 1 ~Br m And Br (Br) 1 ~Br m Sequentially arranged to form a matrix. Second characteristic information B 1 ~B m The matrix corresponding to the second color channel is: bg 1 ~Bg m And Bg 1 ~Bg m Sequentially arranged to form a matrix. Second characteristic information B 1 ~B m The matrix corresponding to the third color channel is: bb 1 ~Bb m And Bb 1 ~Bb m Sequentially arranged to form a matrix.
Then, the remote sensing data platform 10 inputs the feature matrix to the preset defect region identification model 200. FIG. 7 is a schematic diagram of a defect region identification model according to an embodiment of the present application. Referring to FIG. 7, the defective area identification model 200 includes a convolution layer 210, a full join layer 220, and a softmax classification layer 230.
Wherein the convolution layer 210 contains a plurality (e.g., 100) of convolution kernels, such that a corresponding first feature map is generated based on the input feature matrix by the convolution kernels contained by the convolution layer 210.
Then, the convolution layer 210 inputs the generated feature map to the full join layer 220, and the full join layer 220 generates an integrated value from the feature map and inputs the integrated value to the softmax classification layer 230. Thereby outputting identification information related to the defective area by the softmax classification layer 230.
Fig. 8 is a schematic diagram of a convolution kernel of a convolution layer 210 according to an embodiment of the present disclosure. Referring to fig. 8, the convolution kernel of the convolution layer 210 is preferably a matrix of n×3, so that the convolution layer 210 only needs to laterally shift the convolution kernel on the feature matrix when performing the convolution operation. Therefore, the convolution kernel of each convolution layer can completely convolve the information of each column of each feature map in the convolution process, and the integrity of the information of each column is not damaged in the feature extraction process.
Fig. 9 is a schematic diagram of a full connectivity layer 220 and a softmax sorting layer 230 according to an embodiment of the application. Referring to fig. 9, the full connection layer 220 includes 2m units, one image block for every two units. For example, unit S 1,1 Sum unit S 1,2 Corresponding to the first image block Z 1 And unit S 1,1 Output is the first image block Z 1 Integration of normal region, element S 1,2 Output is the first image block Z 1 Integration of the defective area; unit S 2,1 Sum unit S 2,2 Corresponding to the second image block Z 2 And unit S 2,1 Output is the second image block Z 2 Integration of normal region, element S 2,2 Output is the second image block Z 2 Integration of the defective area; unit S m,1 Sum unit S m,2 Corresponding to the mth image block Z m And unit S m,1 Output is the mth image block Z m Integration of normal region, element S m,2 Output is the mth image block Z m Integration of the defect area.
In addition, the softmax classification layer 230 of the defect region identification model 200 includes m softmax classifiers (i.e., softmax 1-softmax m). Wherein each softmax classifier corresponds to a different image block.
For example, softmax 1 is associated with the first image block Z 1 Correspondingly, the unit S receiving the full connection layer 220 1,1 Sum unit S 1,2 Integration of the output and output with the first image block Z 1 Corresponding recognition result Q 1,1 And Q 1,2 . Wherein Q is 1,1 For indicating the first image block Z 1 Probability of being normal region, Q 1,2 For indicating the first image block Z 1 Probability of being a defective area.
Similarly, softmax m and mth image block Z m Correspondingly, the unit S receiving the full connection layer 220 m,1 Sum unit S m,2 Integration of the output, and output with the mth image block Z m Corresponding recognition result Q m,1 And Q m,2 . Wherein Q is m,1 For indicating the mth image block Z m Probability of being normal region, Q m,2 For indicating the mth image block Z m Probability of being a defective area.
Thus, the remote sensing data platform 10 can determine which image blocks belong to the defect area and which image blocks belong to the normal area according to the output result of the softmax classification layer 230.
Optionally, the operation of generating the second feature information corresponding to the first feature information according to the correlation between the first feature information corresponding to each image block includes: selecting a first characteristic information as a target object from a plurality of first characteristic information; determining a correlation coefficient between first characteristic information as a target object and other first characteristic information; and generating second characteristic information respectively corresponding to the first characteristic information according to the correlation coefficient.
Specifically, first, the telemetry data platform 10 needs to select a plurality of first feature information A 1 ~A m The first characteristic information specified in (i.e., the first characteristic information as the target object). For example, as the first characteristic information A 1 Remote sensing numberAccording to the first image block Z 1 Corresponding first characteristic information A 1 Respectively with the first characteristic information A 1 ~A m Correlation with first characteristic information A is generated 1 Corresponding second characteristic information B 1 . Is designated as first characteristic information A 2 The remote sensing data platform 10 then generates a second image block Z 2 Corresponding first characteristic information A 2 Respectively with the first characteristic information A 1 ~A m Correlation with first characteristic information A is generated 2 Corresponding second characteristic information B 2 . Is designated as first characteristic information A m The remote sensing data platform 10 generates a second image block Z according to the second image block Z m Corresponding first characteristic information A m Respectively with the first characteristic information A 1 ~A m Correlation with first characteristic information A is generated m Corresponding second characteristic information B m 。
The telemetry data platform 10 then determines a correlation coefficient between the specified first characteristic information and the other first characteristic information.
Below to the first image block Z 1 Corresponding first characteristic information A 1 An example is described. Specifically, the remote sensing data platform 10 determines the first characteristic information a 1 And first characteristic information A 1 Correlation coefficient R between 1,1 The method comprises the steps of carrying out a first treatment on the surface of the The remote sensing data platform 10 determines the first characteristic information a 1 And first characteristic information A 2 Correlation coefficient R between 1,2 The method comprises the steps of carrying out a first treatment on the surface of the And so on; the remote sensing data platform 10 determines the first characteristic information a 1 And first characteristic information A m Correlation coefficient R between 1,m . Wherein, the correlation coefficient is used for reflecting the height of the correlation. That is, the larger the correlation coefficient is, the higher the correlation is explained; the smaller the correlation coefficient, the lower the correlation is explained.
Due to the respective image blocks Z 1 ~Z m First characteristic information a of (a) 1 ~A m Each including feature vectors corresponding to three color channels (i.e., color channel RGB). Thus the correlation coefficient R 1,1 ~R 1,m The expression is as follows: r is R 1,1 =[Rr 1,1 ,Rg 1,1 ,Rb 1,1 ] T ，R 1,2 =[Rr 1,2 ,Rg 1,2 ,Rb 1,2 ] T Similarly, R 1,m =[Rr 1,m ,Rg 1,m ,Rb 1,m ] T 。
Wherein Rr 1,i Representing first characteristic information A 1 Feature vector Ar of (2) 1 And first characteristic information A i Feature vector Ar of (2) i Correlation coefficient between the two. Rg 1,i Representing first characteristic information A 1 Is of the eigenvector Ag of (2) 1 And first characteristic information A i Is of the eigenvector Ag of (2) i Correlation coefficient between the two. Rb (Rb) 1,i Representing first characteristic information A 1 Feature vector Ab of (2) 1 And first characteristic information A i Feature vector Ab of (2) i Correlation coefficient between the two. Where i= 1~m.
The telemetry data platform 10 then uses the softmax function and based on the correlation coefficient R 1,1 ~R 1,m Determining the correlation coefficients R respectively 1,1 ~R 1,m Corresponding weight value W 1,1 ~W 1,m . Wherein W is 1,1 =[wr 1,1 ,wg 1,1 ,wb 1,1 ] T ，W 1,2 =[wr 1,2 ,wg 1,2 ,wb 1,2 ] T Similarly, W 1,m =[wr 1,m ,wg 1,m ,wb 1,m ] T . Wherein wr is 1,i Representing first characteristic information A 1 Feature vector Ar of (2) 1 And first characteristic information A i Feature vector Ar of (2) i And a weight therebetween. wg (wg) 1,i Representing first characteristic information A 1 Is of the eigenvector Ag of (2) 1 And first characteristic information A i Is of the eigenvector Ag of (2) i And (5) weighting. wb (Wb) 1,i Representing first characteristic information A 1 Feature vector Ab of (2) 1 And first characteristic information A i Feature vector Ab of (2) i And (5) weighting. Where i= 1~m.
Then, the remote sensing data platform 10 calculates the first characteristic information A according to the following formula 1 Corresponding second characteristic information B 1 =[Br 1 ,Bg 1 ,Bb 1 ]：
Referring to the above procedure, for the and image block Z 2 ~Z m Corresponding first characteristic information A 2 ~A m Respectively generating corresponding second characteristic information B 2 ~B m 。
The following still follows with the first characteristic information A 1 For example, calculate the correlation coefficient R 1,1 ~R 1,m Is a process of (2).
Specifically, for the first feature information a as the specified object 1 With other first characteristic information A than the specified object i (where i= 1~m), first feature information a as a specified object is calculated by the following formula 1 With other first characteristic information A than the specified object i Correlation coefficient R between 1,i =[Rr 1,i ,Rg 1，i ,Rb 1,i ]：
Similarly, for other first characteristic information A k (where k= 2~m), the first characteristic information a is also calculated with reference to the above 1 And first characteristic information A i Correlation coefficient R between k,i Is carried out by a method comprising the steps of (a) and (b). That is, the first characteristic information A is calculated k And first characteristic information A i Correlation coefficient R between k,i =[Rr k,i ,Rg k，i ,Rb k,i ]：
Therefore, by determining the correlation coefficient between the first feature information serving as the target object and the other first feature information and generating the second feature information corresponding to the first feature information according to the correlation coefficient, the technical effect of providing the necessary condition for the remote sensing data platform 10 to determine whether the target forest has a defect (i.e., whether the target forest is a defect area) according to the second feature information is achieved.
Optionally, the operation of generating the first feature information corresponding to each image block according to the pixel value of each image block includes: determining a plurality of color channels corresponding to pixel values of respective image blocks; and sequentially arranging the plurality of color channels to generate first characteristic information corresponding to each image block.
Specifically, with reference to the above, each image block Z 1 ~Z m Contains a corresponding plurality of color channels. Thus for image block Z 1 ~Z m Each image block in the plurality of color channels is connected with each row in sequence from head to tail to form a plurality of image blocks Z 1 ~Z m Corresponding first characteristic information A 1 ~A m Reference is made to fig. 5A, 5B and 5C. Furthermore, each image block Z 1 ~Z m First characteristic information a of (a) 1 ~A m Each containing feature vectors corresponding to a different color channel. For example, a first image block Z 1 First characteristic information a of (a) 1 The method comprises the following steps of: ar (Ar) 1 、Ag 1 Ab 1 The method comprises the steps of carrying out a first treatment on the surface of the Second image block Z 2 First characteristic information a of (a) 2 The method comprises the following steps of: ar (Ar) 2 、Ag 2 Ab 2 The method comprises the steps of carrying out a first treatment on the surface of the Third image block Z 3 First characteristic information a of (a) 3 The method comprises the following steps of: ar (Ar) 3 、Ag 3 Ab 3 Similarly, the mth image block Z m First characteristic information a of (a) m The method comprises the following steps of: ar (Ar) m 、Ag m Ab m 。
Further, according to a third aspect of the present embodiment, there is provided a storage medium. The storage medium includes a stored program, wherein the method of any one of the above is performed by a processor when the program is run.
Since the specified image block in the present application has a correlation with other image blocks than the specified image block, the second feature information includes not only the image features of the specified image block (for example, the image features of the image block corresponding to the defective area) but also the image features of the other image blocks than the specified image block (for example, the image features of the image block corresponding to the normal area). Further, since the present application determines whether or not the target forest has a defect based on the second feature information, the difference between the image block of the defective area and the image block of the normal area can be better presented by comparing the second feature information of the designated image block (for example, the image feature of the image block corresponding to the defective area) with the second feature information of the peculiar image block other than the designated image block (for example, the image feature of the image block corresponding to the normal area). Therefore, the technical effect of accurately identifying the defect area in the forest is achieved through the operation. The method solves the technical problems that in the prior art, as the existing method of inputting the remote sensing image into the convolutional neural network so as to identify the defective area in the forest, the difference characteristics between the defective area and the normal area are difficult to excavate, and the defective area in the forest is difficult to accurately identify.
It should be noted that, for simplicity of description, the foregoing method embodiments are all described as a series of acts, but it should be understood by those skilled in the art that the present invention is not limited by the order of acts described, as some steps may be performed in other orders or concurrently in accordance with the present invention. Further, those skilled in the art will also appreciate that the embodiments described in the specification are all preferred embodiments, and that the acts and modules referred to are not necessarily required for the present invention.
From the description of the above embodiments, it will be clear to a person skilled in the art that the method according to the above embodiments may be implemented by means of software plus the necessary general hardware platform, but of course also by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. ROM/RAM, magnetic disk, optical disk) comprising instructions for causing a terminal device (which may be a mobile phone, a computer, a server, or a network device, etc.) to perform the method according to the embodiments of the present invention.
Example 2
Fig. 10 shows a remote sensing image-based forest defect area detection apparatus 1000 according to the first aspect of the present embodiment, the apparatus 1000 corresponding to the method according to the first aspect of embodiment 1. Referring to fig. 10, the apparatus 1000 includes: the location information determining module 1010 is configured to determine location information of a target forest, and determine a remote sensing image corresponding to the target forest according to the location information of the target forest; a forest image area determining module 1020, configured to determine a forest image area corresponding to the target forest in the remote sensing image; the image block dividing module 1030 is configured to divide the forest image area into a plurality of image blocks with the same size, where each image block includes a corresponding pixel value; a first feature information generating module 1040, configured to generate first feature information corresponding to each image block according to a pixel value of each image block, where the first feature information is used to indicate an image feature of the corresponding image block; a second feature information generating module 1050, configured to generate second feature information corresponding to the first feature information according to the correlation between the first feature information corresponding to each image block, where the second feature information is used to indicate image features of all the image blocks; and a determining module 1060, configured to determine whether the target forest has a defect according to the second feature information.
Optionally, the determining module 1060 includes: the model determining module is used for determining a defect area identification model; and the judging submodule is used for inputting the second characteristic information into the defect area identification model and judging whether the target forest has defects or not.
Optionally, the second feature information generating module 1050 includes: the first characteristic information selecting module is used for selecting first characteristic information serving as a target object from a plurality of first characteristic information; a correlation coefficient determination module for determining a correlation coefficient between first feature information as a target object and other first feature information; and the second characteristic information generation sub-module is used for generating second characteristic information corresponding to the first characteristic information respectively according to the correlation coefficient.
Optionally, the first feature information generating module 1040 includes: a color channel determining module for determining a plurality of color channels corresponding to pixel values of respective image blocks; and the first characteristic information generation sub-module is used for sequentially arranging a plurality of color channels to generate first characteristic information corresponding to each image block.
Since the specified image block in the present application has a correlation with other image blocks than the specified image block, the second feature information includes not only the image features of the specified image block (for example, the image features of the image block corresponding to the defective area) but also the image features of the other image blocks than the specified image block (for example, the image features of the image block corresponding to the normal area). Further, since the present application determines whether or not the target forest has a defect based on the second feature information, the difference between the image block of the defective area and the image block of the normal area can be better presented by comparing the second feature information of the designated image block (for example, the image feature of the image block corresponding to the defective area) with the second feature information of the peculiar image block other than the designated image block (for example, the image feature of the image block corresponding to the normal area). Therefore, the technical effect of accurately identifying the defect area in the forest is achieved through the operation. The method solves the technical problems that in the prior art, as the existing method of inputting the remote sensing image into the convolutional neural network so as to identify the defective area in the forest, the difference characteristics between the defective area and the normal area are difficult to excavate, and the defective area in the forest is difficult to accurately identify.
Example 3
Fig. 11 shows a remote sensing image-based forest defect area detection apparatus 1100 according to the first aspect of the present embodiment, the apparatus 1100 corresponding to the method according to the first aspect of embodiment 1. Referring to fig. 11, the apparatus 1100 includes: a processor 1110; and a memory 1120 coupled to the processor 1110 for providing instructions to the processor 1110 for processing the following processing steps: determining the position information of the target forest, and determining a remote sensing image corresponding to the target forest according to the position information of the target forest; in the remote sensing image, determining a forest image area corresponding to a target forest; dividing a forest image area into a plurality of image blocks with the same size, wherein each image block contains a corresponding pixel value; generating first characteristic information corresponding to each image block according to the pixel value of each image block, wherein the first characteristic information is used for indicating the image characteristics of the corresponding image block; generating second characteristic information corresponding to the first characteristic information according to the correlation between the first characteristic information corresponding to each image block, wherein the second characteristic information is used for indicating the image characteristics of all the image blocks; and judging whether the target forest has defects according to the second characteristic information.
Since the specified image block in the present application has a correlation with other image blocks than the specified image block, the second feature information includes not only the image features of the specified image block (for example, the image features of the image block corresponding to the defective area) but also the image features of the other image blocks than the specified image block (for example, the image features of the image block corresponding to the normal area). Further, since the present application determines whether or not the target forest has a defect based on the second feature information, the difference between the image block of the defective area and the image block of the normal area can be better presented by comparing the second feature information of the designated image block (for example, the image feature of the image block corresponding to the defective area) with the second feature information of the peculiar image block other than the designated image block (for example, the image feature of the image block corresponding to the normal area). Therefore, the technical effect of accurately identifying the defect area in the forest is achieved through the operation. The method solves the technical problems that in the prior art, as the existing method of inputting the remote sensing image into the convolutional neural network so as to identify the defective area in the forest, the difference characteristics between the defective area and the normal area are difficult to excavate, and the defective area in the forest is difficult to accurately identify.
The foregoing embodiment numbers of the present invention are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments.
In the foregoing embodiments of the present invention, the descriptions of the embodiments are emphasized, and for a portion of this disclosure that is not described in detail in this embodiment, reference is made to the related descriptions of other embodiments.
In the several embodiments provided in the present application, it should be understood that the disclosed technology content may be implemented in other manners. The above-described embodiments of the apparatus are merely exemplary, and the division of the units, such as the division of the units, is merely a logical function division, and may be implemented in another manner, for example, multiple units or components may be combined or may be integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be through some interfaces, units or modules, or may be in electrical or other forms.
The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied essentially or in part or all of the technical solution or in part in the form of a software product stored in a storage medium, including instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a removable hard disk, a magnetic disk, or an optical disk, or other various media capable of storing program codes.
The foregoing is merely a preferred embodiment of the present invention and it should be noted that modifications and adaptations to those skilled in the art may be made without departing from the principles of the present invention, which are intended to be comprehended within the scope of the present invention.
Claims (10)
1. The forest defect area detection method based on the remote sensing image is characterized by comprising the following steps of:
determining the position information of a target forest, and determining a remote sensing image corresponding to the target forest according to the position information of the target forest;
determining a forest image area corresponding to the target forest in the remote sensing image;
dividing the forest image area into a plurality of image blocks with the same size, wherein each image block contains a corresponding pixel value;
generating first characteristic information corresponding to each image block according to the pixel value of each image block, wherein the first characteristic information is used for representing the image characteristics of the appointed image block;
generating second characteristic information corresponding to the first characteristic information according to the correlation between the first characteristic information corresponding to each image block, wherein the second characteristic information is used for indicating the image characteristics of all the image blocks; and
And judging whether the target forest has defects or not according to the second characteristic information.
2. The method of claim 1, wherein determining whether the target forest is defective based on the second characteristic information comprises:
determining a defect area identification model; and
and inputting the second characteristic information into the defect area identification model, and judging whether the target forest has defects or not.
3. The method according to claim 2, wherein the operation of generating second feature information corresponding to the first feature information, respectively, from correlations between the first feature information corresponding to the respective image blocks, comprises:
selecting a first characteristic information as a target object from a plurality of first characteristic information;
determining a correlation coefficient between first characteristic information as a target object and other first characteristic information; and
and generating second characteristic information corresponding to the first characteristic information respectively according to the correlation coefficient.
4. A method according to claim 3, wherein generating first characteristic information corresponding to the respective image block from pixel values of the respective image block comprises:
Determining a plurality of color channels corresponding to pixel values of the respective image blocks; and
and sequentially arranging the color channels to generate first characteristic information corresponding to each image block.
5. A storage medium comprising a stored program, wherein the method of any one of claims 1 to 4 is performed by a processor when the program is run.
6. Forest defect area detection device based on remote sensing image, characterized by comprising:
the position information determining module is used for determining the position information of the target forest and determining a remote sensing image corresponding to the target forest according to the position information of the target forest;
the forest image area determining module is used for determining a forest image area corresponding to the target forest in the remote sensing image;
the image block dividing module is used for dividing the forest image area into a plurality of image blocks with the same size, wherein each image block contains a corresponding pixel value;
the first characteristic information generation module is used for generating first characteristic information corresponding to each image block according to the pixel value of each image block, wherein the first characteristic information is used for indicating the image characteristic of the corresponding image block;
A second feature information generating module, configured to generate second feature information corresponding to the first feature information according to correlation between the first feature information corresponding to each image block, where the second feature information is used to indicate image features of all image blocks; and
and the judging module is used for judging whether the target forest has defects according to the second characteristic information.
7. The apparatus of claim 6, wherein the means for determining comprises:
the model determining module is used for determining a defect area identification model; and
and the judging submodule is used for inputting the second characteristic information into the defect area identification model and judging whether the target forest has defects or not.
8. The apparatus of claim 7, wherein the second characteristic information generating module comprises:
the first characteristic information selecting module is used for selecting first characteristic information serving as a target object from a plurality of first characteristic information;
a correlation coefficient determination module for determining a correlation coefficient between first feature information as a target object and other first feature information; and
and the second characteristic information generation sub-module is used for generating second characteristic information corresponding to the first characteristic information respectively according to the correlation coefficient.
9. The apparatus of claim 8, wherein the first characteristic information generation module comprises:
a color channel determining module, configured to determine a plurality of color channels corresponding to pixel values of the respective image blocks; and
and the first characteristic information generation sub-module is used for sequentially arranging the plurality of color channels to generate first characteristic information corresponding to each image block.
10. Forest defect area detection device based on remote sensing image, characterized by comprising:
a processor; and
a memory, coupled to the processor, for providing instructions to the processor to process the following processing steps:
determining the position information of a target forest, and determining a remote sensing image corresponding to the target forest according to the position information of the target forest;
determining a forest image area corresponding to the target forest in the remote sensing image;
dividing the forest image area into a plurality of image blocks with the same size, wherein each image block contains a corresponding pixel value;
generating first characteristic information corresponding to each image block according to the pixel value of each image block, wherein the first characteristic information is used for indicating the image characteristic of the corresponding image block;
Generating second characteristic information corresponding to the first characteristic information according to the correlation between the first characteristic information corresponding to each image block, wherein the second characteristic information is used for indicating the image characteristics of all the image blocks; and
and judging whether the target forest has defects or not according to the second characteristic information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310261889.8A CN116385375B (en) | 2023-03-17 | 2023-03-17 | Forest defect area detection method and device based on remote sensing image and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310261889.8A CN116385375B (en) | 2023-03-17 | 2023-03-17 | Forest defect area detection method and device based on remote sensing image and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116385375A true CN116385375A (en) | 2023-07-04 |
| CN116385375B CN116385375B (en) | 2023-10-20 |
Family
ID=86972330
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310261889.8A Active CN116385375B (en) | 2023-03-17 | 2023-03-17 | Forest defect area detection method and device based on remote sensing image and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116385375B (en) |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106296689A (en) * | 2016-08-10 | 2017-01-04 | 常州信息职业技术学院 | Flaw detection method, system and device |
| CN109523533A (en) * | 2018-11-14 | 2019-03-26 | 北京奇艺世纪科技有限公司 | A kind of image quality evaluating method and device |
| US20190220972A1 (en) * | 2018-01-17 | 2019-07-18 | Tokyo Electron Limited | Substrate defect inspection apparatus, substrate defect inspection method, and storage medium |
| CN110675376A (en) * | 2019-09-20 | 2020-01-10 | 福建工程学院 | PCB defect detection method based on template matching |
| CN111738735A (en) * | 2020-07-23 | 2020-10-02 | 腾讯科技（深圳）有限公司 | Image data processing method and device and related equipment |
| CN113205512A (en) * | 2021-05-26 | 2021-08-03 | 北京市商汤科技开发有限公司 | Image anomaly detection method, device, equipment and computer readable storage medium |
| US20210304035A1 (en) * | 2020-03-31 | 2021-09-30 | Panasonic Intellectual Property Management Co., Ltd. | Method and system to detect undefined anomalies in processes |
| CN114022375A (en) * | 2021-11-03 | 2022-02-08 | 深圳绿米联创科技有限公司 | Image processing method, image processing device, electronic equipment and computer readable storage medium |
| CN114331949A (en) * | 2021-09-29 | 2022-04-12 | 腾讯科技（上海）有限公司 | Image data processing method, computer equipment and readable storage medium |
| CN114445318A (en) * | 2020-10-16 | 2022-05-06 | 合肥欣奕华智能机器有限公司 | Defect detection method and device, electronic equipment and storage medium |
| CN115082490A (en) * | 2022-08-23 | 2022-09-20 | 腾讯科技（深圳）有限公司 | Anomaly prediction method, and training method, device and equipment of anomaly prediction model |
-
2023
- 2023-03-17 CN CN202310261889.8A patent/CN116385375B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106296689A (en) * | 2016-08-10 | 2017-01-04 | 常州信息职业技术学院 | Flaw detection method, system and device |
| US20190220972A1 (en) * | 2018-01-17 | 2019-07-18 | Tokyo Electron Limited | Substrate defect inspection apparatus, substrate defect inspection method, and storage medium |
| CN109523533A (en) * | 2018-11-14 | 2019-03-26 | 北京奇艺世纪科技有限公司 | A kind of image quality evaluating method and device |
| CN110675376A (en) * | 2019-09-20 | 2020-01-10 | 福建工程学院 | PCB defect detection method based on template matching |
| US20210304035A1 (en) * | 2020-03-31 | 2021-09-30 | Panasonic Intellectual Property Management Co., Ltd. | Method and system to detect undefined anomalies in processes |
| CN111738735A (en) * | 2020-07-23 | 2020-10-02 | 腾讯科技（深圳）有限公司 | Image data processing method and device and related equipment |
| CN114445318A (en) * | 2020-10-16 | 2022-05-06 | 合肥欣奕华智能机器有限公司 | Defect detection method and device, electronic equipment and storage medium |
| CN113205512A (en) * | 2021-05-26 | 2021-08-03 | 北京市商汤科技开发有限公司 | Image anomaly detection method, device, equipment and computer readable storage medium |
| CN114331949A (en) * | 2021-09-29 | 2022-04-12 | 腾讯科技（上海）有限公司 | Image data processing method, computer equipment and readable storage medium |
| CN114022375A (en) * | 2021-11-03 | 2022-02-08 | 深圳绿米联创科技有限公司 | Image processing method, image processing device, electronic equipment and computer readable storage medium |
| CN115082490A (en) * | 2022-08-23 | 2022-09-20 | 腾讯科技（深圳）有限公司 | Anomaly prediction method, and training method, device and equipment of anomaly prediction model |
Non-Patent Citations (2)
| Title |
|---|
| AQSA RASHEED等: "Fabric Defect Detection Using Computer Vision Techniques: A Comprehensive Review", MATHEMATICAL PROBLEMS IN ENGINEERING, pages 1 - 24 * |
| 马逐曦: "基于超像素的平面铣削工件表面缺陷视觉检测研究", 中国优秀硕士学位论文全文数据库工程科技Ⅰ辑, no. 02, pages 022 - 895 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116385375B (en) | 2023-10-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109584248B (en) | Infrared target instance segmentation method based on feature fusion and dense connection network | |
| US9047678B2 (en) | Change analyst | |
| CN112154451A (en) | Method, apparatus and computer program for extracting representative features of objects in an image | |
| CN112633185B (en) | Image processing method and device | |
| CN107480706A (en) | A kind of seed production corn field remote sensing recognition method and device | |
| EP1700269A2 (en) | Detection of sky in digital color images | |
| CN113159300A (en) | Image detection neural network model, training method thereof and image detection method | |
| CN113988147B (en) | Multi-label classification method and device for remote sensing image scene based on graph network, and multi-label retrieval method and device | |
| CN109284741A (en) | A kind of extensive Remote Sensing Image Retrieval method and system based on depth Hash network | |
| CN101300576A (en) | Image comparison | |
| CN113449806A (en) | Two-stage forestry pest identification and detection system and method based on hierarchical structure | |
| CN114494821A (en) | Remote sensing image cloud detection method based on feature multi-scale perception and self-adaptive aggregation | |
| CN114742733A (en) | Cloud removing method and device, computer equipment and storage medium | |
| CN115375548A (en) | Super-resolution remote sensing image generation method, system, equipment and medium | |
| CN112686184A (en) | Remote sensing house change detection method based on neural network | |
| Sankaran et al. | Can High-Resolution Satellite Multispectral Imagery Be Used to Phenotype Canopy Traits and Yield Potential in Field Conditions? | |
| CN116385375B (en) | Forest defect area detection method and device based on remote sensing image and storage medium | |
| CN111476739B (en) | Underwater image enhancement method, system and storage medium | |
| CN109740432A (en) | Plant monitoring method, device, computer equipment and storage medium | |
| Mahule et al. | Hybrid Method for Improving Accuracy of Crop-Type Detection using Machine Learning | |
| CN111274145A (en) | Relationship structure chart generation method and device, computer equipment and storage medium | |
| CN111401453A (en) | Mosaic image classification and identification method and system | |
| CN116092019A (en) | Ship periphery abnormal object monitoring system, storage medium thereof and electronic equipment | |
| Shahtahmassebi et al. | Evaluation on the two filling functions for the recovery of forest information in mountainous shadows on Landsat ETM+ Image | |
| CN114494887A (en) | Remote sensing image classification method and device, computer equipment and readable storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |