CN110175503A - Length acquisition methods, device, settlement of insurance claim system, medium and electronic equipment - Google Patents
Length acquisition methods, device, settlement of insurance claim system, medium and electronic equipment Download PDFInfo
- Publication number
- CN110175503A CN110175503A CN201910271246.5A CN201910271246A CN110175503A CN 110175503 A CN110175503 A CN 110175503A CN 201910271246 A CN201910271246 A CN 201910271246A CN 110175503 A CN110175503 A CN 110175503A
- Authority
- CN
- China
- Prior art keywords
- length
- object area
- area
- pixels
- image
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 65
- 238000003709 image segmentation Methods 0.000 claims abstract description 46
- 230000011218 segmentation Effects 0.000 claims abstract description 20
- 238000003860 storage Methods 0.000 claims abstract description 10
- 230000006870 function Effects 0.000 claims description 53
- 244000144972 livestock Species 0.000 claims description 42
- 238000012549 training Methods 0.000 claims description 35
- 238000012937 correction Methods 0.000 claims description 23
- 238000012545 processing Methods 0.000 claims description 19
- 238000012360 testing method Methods 0.000 claims description 16
- 238000004590 computer program Methods 0.000 claims description 12
- 238000001514 detection method Methods 0.000 claims description 12
- 238000000605 extraction Methods 0.000 claims description 8
- 230000004927 fusion Effects 0.000 claims description 5
- 230000004044 response Effects 0.000 claims description 5
- 230000000977 initiatory effect Effects 0.000 claims description 4
- 238000012935 Averaging Methods 0.000 claims description 3
- 241001465754 Metazoa Species 0.000 claims description 3
- 244000144977 poultry Species 0.000 claims description 3
- 230000001419 dependent effect Effects 0.000 abstract description 4
- 241000282898 Sus scrofa Species 0.000 description 42
- 238000010586 diagram Methods 0.000 description 12
- 210000003414 extremity Anatomy 0.000 description 11
- 230000008569 process Effects 0.000 description 8
- 238000004364 calculation method Methods 0.000 description 7
- 238000005259 measurement Methods 0.000 description 7
- 238000004891 communication Methods 0.000 description 5
- 241000282887 Suidae Species 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 4
- 238000003062 neural network model Methods 0.000 description 4
- 241000208340 Araliaceae Species 0.000 description 3
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 3
- 235000003140 Panax quinquefolius Nutrition 0.000 description 3
- 230000004913 activation Effects 0.000 description 3
- 210000003194 forelimb Anatomy 0.000 description 3
- 235000008434 ginseng Nutrition 0.000 description 3
- 230000003993 interaction Effects 0.000 description 3
- 210000003141 lower extremity Anatomy 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 239000011248 coating agent Substances 0.000 description 2
- 238000000576 coating method Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 230000001755 vocal effect Effects 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
- 230000004584 weight gain Effects 0.000 description 1
- 235000019786 weight gain Nutrition 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/08—Insurance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
-
- 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
- G06T2207/10004—Still image; Photographic image
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Development Economics (AREA)
- Technology Law (AREA)
- General Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Economics (AREA)
- Geometry (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The embodiment of the invention discloses a kind of length acquisition methods, device, settlement of insurance claim system, storage medium and electronic equipment, the length acquisition methods include: to obtain the image comprising object and object of reference；Based on Image Segmentation Model, segmentation obtains target object area and referring to object area from described image；The physical length of the length in pixels referring to object area and the object of reference is obtained, and obtains the length in pixels of the target object area；In conjunction with the physical length of the length in pixels referring to object area and the object of reference, the physical length of the object is obtained according to the length in pixels of the target object area.Solve the problems, such as that linear measure longimetry in the prior art is realized dependent on artificial using length acquisition methods provided in an embodiment of the present invention, device, settlement of insurance claim system, storage medium and electronic equipment.
Description
Technical field
The present invention relates to computer field more particularly to a kind of length acquisition methods, device, settlement of insurance claim system, storages
Medium and electronic equipment.
Background technique
Linear measure longimetry has gradually been applied in various fields, for example, in the settlement of insurance claim for livestock, since livestock body is long
It is the important indicator for reflecting livestock upgrowth situation, therefore, it is long to determine that the Claims Resolution scheme of insurance company is normally based on livestock body
Amount for which loss settled.
Currently, linear measure longimetry is mainly the method for taking artificial detection object length, specifically, survey crew passes through volume
The measuring tools such as ruler from first to last or from top to bottom measure object.
However, not only inefficiency, measurement are tied inventors realized that the method process is cumbersome and dependent on manually realizing
Fruit depends greatly on the subjectivity judgement of survey crew, it is difficult to guarantee the accuracy of measurement result.
From the foregoing, it will be observed that how to avoid linear measure longimetry still urgently to be resolved dependent on artificial realization.
Summary of the invention
The embodiment of the present invention provides a kind of length acquisition methods, device, settlement of insurance claim system, electronic equipment and storage and is situated between
Matter can solve linear measure longimetry present in the relevant technologies dependent on artificial the problem of realizing.
Wherein, the technical scheme adopted by the invention is as follows:
One side according to an embodiment of the present invention, a kind of length acquisition methods, comprising: obtaining includes object and object of reference
Image；Based on Image Segmentation Model, segmentation obtains target object area and referring to object area from described image；Obtain the ginseng
The physical length of length in pixels and the object of reference according to object area, and obtain the length in pixels of the target object area；Knot
The physical length for closing the length in pixels referring to object area and the object of reference, according to the length in pixels of the target object area
Obtain the physical length of the object.
One side according to an embodiment of the present invention, a kind of length acquisition device, comprising: image collection module, for obtaining
Image comprising object and object of reference；Image segmentation module is divided from described image for being based on Image Segmentation Model
To target object area and referring to object area；Length in pixels computing module, for obtain the length in pixels referring to object area with
The physical length of the object of reference, and obtain the length in pixels of the target object area；Object length computation module, is used for
In conjunction with the physical length of the length in pixels referring to object area and the object of reference, the pixel according to the target object area is long
Degree calculates the physical length of the object.
One side according to an embodiment of the present invention, a kind of settlement of insurance claim system, including request end, Claims Resolution end and test side,
Wherein, the request end, for initiating settlement of insurance claim request to the Claims Resolution end, carrying in the settlement of insurance claim request includes domestic animal
The image of poultry and object of reference；Described image is sent to the inspection for requesting in response to the settlement of insurance claim by the Claims Resolution end
Survey end；The test side, for carrying out the long detection of livestock body according to described image, the long detection of the livestock body includes: based on figure
As parted pattern, segmentation obtains livestock region and referring to object area, obtains mesh by the livestock region disconnecting from described image
Object area is marked, and obtains the length in pixels and the length in pixels referring to object area of the target object area, in conjunction with the ginseng
The physical length of length in pixels and the object of reference according to object area, according to the acquisition of the length in pixels of the target object area
The estimation body of livestock is long；The Claims Resolution end, the estimation body for being also used to receive the livestock that the test side returns is long, and according to
The estimation body of the livestock is long to carry out settlement of insurance claim.
One side according to an embodiment of the present invention, a kind of electronic equipment, including processor and memory, on the memory
It is stored with computer program, the computer program realizes length acquisition methods as described above when being executed by the processor.
One side according to an embodiment of the present invention, a kind of storage medium are stored thereon with computer program, the computer
Length acquisition methods as described above are realized when program is executed by processor.
In the above-mentioned technical solutions, the image comprising object and object of reference is obtained, is based on Image Segmentation Model from image
Segmentation obtains target object area and referring to object area, with calculate the length in pixels of target object area and referring to the pixel of object area it is long
Degree, and then the physical length of object of reference is combined, the physical length of object is obtained according to the length in pixels of target object area, by
This, the length of object is detected by the object of reference in conjunction with Image Segmentation Model and where being placed in object in space automatically
Degree efficiently solves the problems, such as that linear measure longimetry in the prior art relies on artificial realize.
It should be understood that above general description and following detailed description be only it is exemplary and explanatory, not
It can the limitation present invention.
Detailed description of the invention
The drawings herein are incorporated into the specification and forms part of this specification, and shows and meets implementation of the invention
Example, and in specification together principle for explaining the present invention.
Fig. 1 is the schematic diagram of related implementation environment according to the present invention.
Fig. 2 is a kind of hardware block diagram of server shown according to an exemplary embodiment.
Fig. 3 is a kind of flow chart of length acquisition methods shown according to an exemplary embodiment.
Fig. 4 is the schematic diagram that livestock region disconnecting involved in Fig. 3 corresponding embodiment goes out target area.
Fig. 5 be in Fig. 3 corresponding embodiment step 370 in the flow chart of one embodiment.
Fig. 6 is a kind of flow chart of length acquisition methods shown according to an exemplary embodiment.
Fig. 7 is a kind of flow chart of length acquisition methods shown according to an exemplary embodiment.
Fig. 8 be Fig. 7 corresponding embodiment in step 530 one embodiment flow chart.
Fig. 9 be in Fig. 3 corresponding embodiment step 330 in the flow chart of one embodiment.
Figure 10 is livestock region and the schematic diagram referring to object area involved in Fig. 9 corresponding embodiment.
Figure 11 is a kind of flow chart of length acquisition methods shown according to an exemplary embodiment.
Figure 12 be in Figure 11 corresponding embodiment step 630 in the flow chart of one embodiment.
Figure 13 be in Fig. 3 corresponding embodiment step 350 in the flow chart of one embodiment.
Figure 14 is the schematic diagram that fitted area involved in Figure 13 corresponding embodiment is rectangle.
Figure 15 is a kind of timing diagram of settlement of insurance claim method in an application scenarios.
Figure 16 is that Figure 15 corresponds to a kind of specific implementation schematic diagram of length acquisition methods in application scenarios.
Figure 17 is a kind of block diagram of length acquisition device shown according to an exemplary embodiment.
Figure 18 is a kind of hardware block diagram of length acquisition device shown according to an exemplary embodiment.
Through the above attached drawings, it has been shown that the specific embodiment of the present invention will be hereinafter described in more detail, these attached drawings
It is not intended to limit the scope of the inventive concept in any manner with verbal description, but is by referring to specific embodiments
Those skilled in the art illustrate idea of the invention.
Specific embodiment
Here will the description is performed on the exemplary embodiment in detail, the example is illustrated in the accompanying drawings.Following description is related to
When attached drawing, unless otherwise indicated, the same numbers in different drawings indicate the same or similar elements.Following exemplary embodiment
Described in embodiment do not represent all embodiments consistented with the present invention.On the contrary, they be only with it is such as appended
The example of device and method being described in detail in claims, some aspects of the invention are consistent.
As previously mentioned, linear measure longimetry still relies on artificial realization, not only inefficiency, it is difficult to guarantee the accurate of linear measure longimetry
Property.
For this purpose, spy of the embodiment of the present invention proposes a kind of length acquisition methods, in conjunction with Image Segmentation Model and it is placed in mesh
Object of reference where marking object in space, carrys out automatic measurement length, can effectively improve the accuracy rate of linear measure longimetry, correspondingly,
It is matched with the length acquisition device of this kind of length acquisition methods, is deployed in the electronic equipment for having von Neumann architecture,
For example, electronic equipment is personal computer (PC), server, smart phone etc., length acquisition methods are realized with this.
Fig. 1 is a kind of schematic diagram of implementation environment involved in length acquisition methods.The implementation environment includes 100 He of terminal
Server end 200.
Wherein, terminal 100 can be desktop computer, laptop, tablet computer, smart phone or other have it is logical
The electronic equipment of linkage function is believed, herein without limiting.
Server end 200 can be a server, be also possible to the server cluster being made of multiple servers, even
It is the cloud computing center being made of multiple servers.This server be the electronic equipment of background service is provided for user, for example, after
Platform service includes but is not limited to linear measure longimetry service etc..
Server end 200 pre-establishes the communication connection between terminal 100 by the modes such as wireless or wired, and leads to
It crosses communication connection and realizes that the data between server end 200 and terminal 100 are transmitted.For example, the data of transmission include image, livestock
Estimation body it is long etc..
By the interaction of terminal 100 and server end 200, user can be asked by terminal 100 to the initiation of server end 200
It asks, so that server end 200 provides linear measure longimetry service.
For server end 200, the request just can be received, and provide a user according to the image in the request
Linear measure longimetry service, it is long with the estimation body for returning to livestock to terminal 100.
Specifically, linear measure longimetry service includes: based on Image Segmentation Model, from the image comprising object and object of reference
Middle segmentation obtains target object area and referring to object area, and with this obtains the length in pixels of target object area and referring to object area
Length in pixels, and then the physical length of object of reference is combined, the actual (tube) length of object is obtained according to the length in pixels of target object area
Degree.
Fig. 2 is a kind of hardware block diagram of server shown according to an exemplary embodiment.This kind of server is applicable in
The server end of the implementation environment shown by Fig. 1.
It should be noted that this kind of server, which is one, adapts to example of the invention, it must not believe that there is provided right
Any restrictions of use scope of the invention.This kind of server can not be construed to need to rely on or must have in Fig. 2
One or more component in illustrative server 200 shown.
The hardware configuration of server 200 can generate biggish difference due to the difference of configuration or performance, as shown in Fig. 2,
Server 200 include: power supply 210, interface 230, at least a memory 250 and an at least central processing unit (CPU,
Central Processing Units)270。
Specifically, power supply 210 is used to provide operating voltage for each hardware device on server 200.
Interface 230 includes an at least wired or wireless network interface, for interacting with external equipment.For example, carrying out Fig. 1 institute
Interaction in implementation environment between terminal 100 and server end 200 is shown.
Certainly, in the example that remaining present invention is adapted to, interface 230 can further include an at least serioparallel exchange and connect
233, at least one input/output interface 235 of mouth and at least USB interface 237 etc., as shown in Fig. 2, herein not to this composition
It is specific to limit.
The carrier that memory 250 is stored as resource, can be read-only memory, random access memory, disk or CD
Deng the resource stored thereon includes operating system 251, application program 253 and data 255 etc., and storage mode can be of short duration
It stores or permanently stores.
Wherein, operating system 251 be used for manage and control server 200 on each hardware device and application program 253,
To realize operation and processing of the central processing unit 270 to mass data 255 in memory 250, Windows can be
ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM etc..
Application program 253 is the computer program based at least one of completion particular job on operating system 251, can
To include an at least module (being not shown in Fig. 2), each module can separately include the series of computation to server 200
Machine readable instruction.For example, length acquisition device can be considered the application program 253 for being deployed in server.
Data 255 can be stored in photo, picture in disk etc., can also be the image comprising livestock and object of reference
Deng being stored in memory 250.
Central processing unit 270 may include the processor of one or more or more, and be set as total by least one communication
Line is communicated with memory 250, to read the computer-readable instruction stored in memory 250, and then is realized in memory 250
The operation and processing of mass data 255.For example, reading the series of computation stored in memory 250 by central processing unit 270
The form of machine readable instruction completes length acquisition methods.
It is appreciated that structure shown in Fig. 2 is only to illustrate, server 200 may also include more more or less than shown in Fig. 2
Component, or with the component different from shown in Fig. 2.Each component shown in Fig. 2 can use hardware, software or its group
It closes to realize.
Referring to Fig. 3, in one exemplary embodiment, a kind of length acquisition methods are suitable for electronic equipment, which sets
It is standby to can be personal computer, smart phone, it can also be the server end of implementation environment shown in Fig. 1, the knot of the electronic equipment
Structure is not limited to the structure of server shown in Fig. 2.
This kind of length acquisition methods can be executed by electronic equipment, may comprise steps of:
Step 310, the image comprising object and object of reference is obtained.
Wherein, where the object of reference can be placed horizontally at the object in space.
For example, using cube as object of reference, it is placed horizontally at the side of livestock for object is livestock, or
Person is placed horizontally at the back of livestock using card as object of reference, and such object of reference is not blocked by livestock, and can be always
It keeps horizontal positioned, effectively reduces inclination, caused ear tag is due to horizontal plane when avoiding with this using ear tag as object of reference
Knockdown and the system perspective error present in imaging.
Further, the size of object of reference is greater than the size of ear tag, in favor of promoting the accuracy rate of the long detection of livestock body.
Image is that Image Acquisition end carries out shooting generation to object and object of reference, which can be cloth
Picture pick-up device where being located at object in space can also be other configurations camera shooting group for example, picture pick-up device is camera
The electronic equipment of part, for example, electronic equipment is the smart phone for being configured with camera.
It, both can be with the figure of real-time image acquisition collection terminal acquisition about the acquisition comprising object and the image of object of reference
Picture, in order in real time to image carry out relevant treatment, also in an available historical time section Image Acquisition end acquire figure
Picture, in order to carry out relevant treatment to image when the task of processing is less, alternatively, to image under the instruction of operator
Relevant treatment is carried out, for example, relevant treatment includes image segmentation, the present embodiment is limited not to this.
In other words, the image comprising object and object of reference got can be adopted in real time from Image Acquisition end
The image of collection can also be the pre-stored image acquired by Image Acquisition end, for example, electronic equipment read by local or
The mode of person's network downloading obtains.
And for Image Acquisition end, the image that electronic equipment is got be acquired by acquisition equipment, either
Image Acquisition end actively uploads acquired image, is also possible to the image issued in response to electronic equipment to Image Acquisition end
Acquisition instruction, and acquired image is uploaded to electronic equipment, it is not also limited herein.
So, electronic equipment can provide linear measure longimetry clothes after getting the image comprising object and object of reference
Business, i.e., by the physical length of object and object of reference calculating object in image.
It certainly,, can also be on Image Acquisition end in order to ensure the accuracy rate of linear measure longimetry for electronic equipment
The image of biography carries out related pretreatment, to remove the noise generated in image transmitting process, wherein correlation pretreatment includes but not
It is limited to the processing such as normalization, filtering.
It is noted that Image Acquisition end shoots object and object of reference, it can be and be continuously shot generation one
Section video can also be that discontinuous shooting generates several pictures, be based on this, and in the embodiment of the present invention, length, which is obtained with frame, is
Unit carries out, and that is to say, image, can refer to the frame video image in one section of video, may also mean that in several pictures
A certain picture, be not limited herein.
Step 330, it is based on Image Segmentation Model, segmentation obtains target object area and referring to object area from described image.
Image segmentation, for being partitioned into target object area from image and referring to object area.Specifically, image segmentation essence
On be to assign the pixel for belonging to the same area in image to identical gray value so that the pixel of different zones has not in image
Same gray value, so that being partitioned into target object area in image and referring to object area.For example, the pixel of target object area
Gray value is 0, and the gray value referring to the pixel of object area is 255.
Wherein, target object area is used to indicate object key position position in image, for example, object key portion
Position includes head, tail portion, four limbs, trunk, ear etc., and correspondingly, target object area includes head zone, tail region, four limbs
Region, torso area, ear region etc..
Object of reference position in image is used to indicate referring to object area.
In the present embodiment, image segmentation is realized based on Image Segmentation Model.
The Image Segmentation Model is with machine learning model for basic model, by training sample to the basic model into
What row model training generated.Wherein, training sample is referred to target object area and the image being labeled referring to object area.
That is, Image Segmentation Model, substantially constructs image and target object area, referring to the number between object area
Mapping relations are learned, then, after an image is obtained, mathe-matical map relationship that can be provided based on Image Segmentation Model, from figure
Segmentation obtains target object area and referring to object area as in.
Optionally, machine learning model includes but is not limited to: convolutional neural networks model, deep neural network model, residual
Poor neural network model etc..
Optionally, image segmentation includes: normal segmentation, semantic segmentation, strength segmentation etc., wherein normal segmentation is further
It include: Threshold segmentation, region segmentation, edge segmentation, histogram divion etc..
In the realization of an embodiment, image segmentation refers to deep neural network model for basic model, to image into
Row semantic segmentation.
Optionally, when object is livestock, the target object area divided from image includes but is not limited to: livestock four
Limb semantic region, carcasses semantic region, livestock ear semantic region and object of reference semantic region.
It should be appreciated that length obtains each application scenarios for being widely used in different field, the practical need of different application scene
Asking may different from.Using object as pig, the length of object is that the long citing of pig body is illustrated, in an application scenarios,
Pig body grows the length for referring to the pig basal part of the ear to pigtail, and in another application scene, pig body grows the length for then referring to pig's head to pigtail,
Correspondingly, target object area is also by different from.
As shown in Fig. 4 (a), in image include head zone 401, ear region 402, torso area 403, four limbs region and
Tail region 405.Wherein, four limbs region includes forelimb region 4041 and hind leg region 4042.
If pig body length refers to the pig basal part of the ear to the length of pigtail, target object area, which refers to, eliminates four limbs region, head
Region, ear region torso area and tail region, as shown in dash area in Fig. 4 (b), and if pig body length refer to pig's head
To the length of pigtail, then target object area is to eliminate four limbs region, the head zone of ear region, torso area and tail portion
Region, as shown in dash area in Fig. 4 (c).
Step 350, the physical length of the length in pixels referring to object area and the object of reference is obtained, and obtains institute
State the length in pixels of target object area.
Firstly, obtain target object area and referring to object area after, can based on target object area and referring to object area,
Determine the length of length and object of reference in the picture of object in the picture.
Wherein, the length of object in the picture, the i.e. length in pixels of target object area.
The length of object of reference in the picture, i.e., referring to the length in pixels of object area.
About length in pixels, can by calculate target object area or referring between key point in object area it is European away from
From indicating, can also indicate by calculating target object area or referring to the length between key point in object area.
For example, key point may come from target object area or the external profile of minimum referring to object area, then, this is most
The length of small external profile can be used as the length in pixels of target object area or the length in pixels referring to object area.
Secondly, the physical length of object of reference, refers to that object of reference is using the measuring tools such as tape measure in space where object
Measure obtained actual measurement length.
It remarks additionally herein, length is substantially to further comprise width, height etc. for the distance of point-to-point
Meaning.
Step 370, in conjunction with the physical length of the length in pixels referring to object area and the object of reference, according to the mesh
The length in pixels of mark object area calculates the physical length of the object.
Determining the length of length and object of reference in the picture of object in the picture, it can be in conjunction with object of reference
Physical length, i.e. object of reference in the actual measurement length obtained in space using the measuring tools measurement such as tape measure where object,
Estimation obtains the physical length that length of the object where it in space is object, herein, since the length is estimation
It obtains, therefore, is defined as the physical length of object, actual length of the object where it in space is different from this.
Specifically, in the realization of an embodiment, as shown in figure 5, step 370 may comprise steps of:
Step 371, multiplying is carried out to the physical length of the length in pixels of the target object area and the object of reference.
Step 373, division arithmetic is carried out to multiplication result and the length in pixels referring to object area, division is transported
Calculate physical length of the result as the object.
It that is to say, the calculation formula of the physical length of object is as follows:
T=(L × Ta)/La。
Wherein, T indicates the physical length of object, and L indicates the physical length of object of reference, TaIndicate object in the picture
Length, i.e. the length in pixels of target object area, LaIndicate the length of object of reference in the picture, i.e., it is long referring to the pixel of object area
Degree.
Length automatic measurement is realized in conjunction with the object of reference of Image Segmentation Model and placement by process as described above,
It not only avoids relying on and is realized in artificial, effectively improve the efficiency of linear measure longimetry, especially can also be kept away when object is livestock
Exempt to cause environmentally-friendly sanitary problem.
Referring to Fig. 6, in one exemplary embodiment, after step 370, method as described above can also include following
Step:
Step 410, the correction factor obtained based on study, zooms in and out processing to the physical length of the object.
It should be appreciated that in Image Acquisition end actual photographed, a possibility that there are system perspective errors when due to imaging, because
This, after obtaining the physical length of object, it is also necessary to make further amendment for the physical length of object.
Amendment is substantially to zoom in and out operation to the physical length of object using correction factor.
Specifically, makeover process uses following calculation formula:
T '=T × (1+r).
Wherein, T ' indicates the physical length of revised object, and T indicates the physical length of object, r indicate correction because
Son.
Step 430, scaling processing result is updated to the physical length of the object.
Under the cooperation of above-described embodiment, the amendment of the physical length of object is realized, avoids having an X-rayed because of system with this
Error and cause the physical length of object to there are problems that error, to further improve the accuracy rate of linear measure longimetry.
It may also be said that significantly reducing target by the modified double shield of the object of reference and correction factor placed
The error of the physical length of object, it is thereby advantageously ensured that the accuracy of linear measure longimetry.
Referring to Fig. 7, in one exemplary embodiment, before step 410, method as described above can also include following
Step:
Step 510, sample to be learned is obtained.
Wherein, the sample to be learned includes the physical length and actual length of object to be learned.
As previously mentioned, the physical length of object, is only the length for estimating obtained object where it in space, area
Actual length not in object where it in space.
The sample to be learned can be and treat learning objective object by Image Acquisition end and object of reference carries out shooting generation and adopts
Collection, similarly, about the acquisition of sample to be learned, the real-time acquisition at Image Acquisition end can be derived from, is also possible to deposit in advance
Storage, details are not described herein.
Step 530, according to the physical length and actual length of object to be learned in the sample to be learned, described in progress
The study of correction factor.
Study, refers to the physical length and actual length according to object to be learned in several samples to be learned, counts
Correction factor is obtained, to be based on correction factor, so that the physical length of object is similar to actual length.
Specifically, in the realization of an embodiment, as shown in figure 8, step 530 may comprise steps of:
Step 531, for each sample to be learned, the physical length of object to be learned in the sample to be learned is obtained
Relative error between actual length.
Step 533, processing of averaging is carried out to the relative error, using the average value as the correction factor.
As an example it is assumed that the physical length of object to be learned is t={ t in sample to be learned1,t2,t3,……,
tn, the actual length of object to be learned is g={ g in sample to be learned1,g2,g3,……,gn, wherein n indicates sample to be learned
This quantity.
Then, the calculation formula of the relative error between the physical length and actual length of object to be learned is as follows:
e1=(t1-g1)/t1,
...,
en=(tn-gn)/tn。
Further, the calculation formula of correction factor is as follows:
R=(e1+e2+e3+……+en)/n。
Wherein, ei(1≤i≤n) indicates relative error.
Under the action of above-described embodiment, the study of correction factor is realized, so that carrying out object based on correction factor
The amendment of physical length be achieved.
Referring to Fig. 9, in one exemplary embodiment, Image Segmentation Model includes input layer, convolutional layer, articulamentum and defeated
Layer out.The convolutional layer is used for feature extraction, which is used for Fusion Features.
Optionally, Image Segmentation Model can also include active coating, pond layer.Wherein, active coating is for improving model instruction
Experienced convergence rate, pond layer is then for reducing the complexity of feature connection.
It remarks additionally, the network topology structure that each level has in Image Segmentation Model is not limited herein
It is fixed, the micronet structure of lightweight can be both used, the efficiency of Lai Tigao image segmentation can also use the network of full dose grade
Structure, fully to ensure the accuracy rate of image segmentation.
Correspondingly, step 330 may comprise steps of:
Step 331, described image is transmitted to convolutional layer by the input layer of described image parted pattern.
Step 333, feature extraction is carried out to described image by the convolutional layer, obtains the local feature of described image.
Step 335, Fusion Features are carried out by local feature of the articulamentum to described image, obtains the overall situation of described image
Feature.
Wherein, local feature includes the shape feature of image, spatial relation characteristics, and global characteristics then include the face of image
Color characteristic, textural characteristics.
That is, the feature extraction through convolutional layer obtains local feature, then the Fusion Features through articulamentum obtain the overall situation
Feature, it is meant that in Image Segmentation Model the feature of different resolution different scale can be interrelated, and simultaneously non-orphaned, with
The accuracy rate of image segmentation is effectively promoted, and then is conducive to improve the accuracy rate of linear measure longimetry.
Step 337, area classification prediction is carried out by global characteristics of the output layer to described image, obtained in described image
Target object area and referring to object area.
In the present embodiment, area classification prediction, is realized by the activation primitive classifier in output layer.It also will be understood that
For the global characteristics based on image belong to different zones classification by each pixel in activation primitive classifier calculated image
Probability, with this realize area classification predict.
Wherein, area classification includes: target object area classification and object of reference area classification.
Specifically, by the activation primitive classifier in output layer, some pixel in the image is calculated and belongs to mesh
The probability for marking object area classification is P1, and it is P2 that pixel, which belongs to the probability of object of reference area classification, in the image.
If P1 > P2, that is, indicate that the pixel belongs to target object area classification in the image, then determines that the pixel is located at target
Object area.
, whereas if P1 < P2, that is, indicate that the pixel belongs to object of reference area classification in the image, then determine the pixel position
In referring to object area.
As a result, after carrying out probability calculation for each pixel in image, each pixel can be determined in image
In region, target object area is realized with this and referring to the segmentation of object area.
Still using object as pig, the length of object is that the long citing of pig body is illustrated, as shown in Figure 10 (a), for packet
For image 600 containing object 601 and object of reference 602, object of reference 602 is a card, is placed horizontally at the back of object 601
Face.
Wherein, target object area is by head zone 401, ear region 402, torso area 403, four limbs region and tailer
Domain 405 forms.Wherein, four limbs region includes forelimb region 4041 and hind leg region 4042, and is 501 referring to object area.
As shown in Figure 10 (b), for the image 600 comprising object 601 and object of reference 603, object of reference 603 is one
Cuboid is placed horizontally at the ground of 601 side of object.
Wherein, target object area is still by head zone 401, ear region 402, torso area 403, four limbs region and tail portion
Region 405 forms.Wherein, four limbs region includes forelimb region 4041 and hind leg region 4042, and is then 502 referring to object area.
It should be noted that in the picture, target object area is removed and the residue referring to the pixel in object area, in image
Pixel constitutes background area.
Please refer to Figure 11, in one exemplary embodiment, method as described above can with the following steps are included:
Step 610, training sample is obtained.
Wherein, the training sample is the image for having carried out target object area mark and object of reference area marking.
Step 630, basic model is constructed, and model training is carried out to the basic model using the training sample, is obtained
To described image parted pattern.
Firstly, basic model, including but not limited to: convolutional neural networks model, deep neural network model, residual error nerve
Network model etc..
Secondly, model training, is to be subject to iteration optimization by parameter of the training sample to basic model, so that by training sample
The assignment algorithm function that the parameter of this and basic model constructs meets the condition of convergence, so that the image of building and object area
Domain is optimal referring to the mathe-matical map relationship between object area, then basic model converges to Image Segmentation Model.
Wherein, assignment algorithm includes but is not limited to: greatest hope function, loss function etc..
Using assignment algorithm as loss function, model training process is illustrated as follows.
Specifically, as shown in figure 12, step 630 may comprise steps of in the realization of an embodiment:
Step 631, based on the parameter for working as previous training sample and the basic model, the damage of the basic model is constructed
Lose function.
Specifically, the loss function of the basic model is the Classification Loss function predicted for area classification, to candidate
Region carries out the loss function of position recurrence, the sum of loss function of introducing separating mask on the candidate region.Wherein, institute
Stating candidate region includes the target object area, the reference object area.
It that is to say, L=Lcls+Lbox+Lmask。
Wherein, L indicates the loss function of basic function, LclsIndicate the Classification Loss function for area classification prediction,
LboxIndicate the loss function that position recurrence is carried out to candidate region, LmaskExpression introduces separating mask on the candidate region
Loss function.
The loss function of loss function building basic model based on different task as a result, just can instruct more accurately
Image Segmentation Model is got, and then is conducive to the accuracy rate of image segmentation, to be conducive to improve the accuracy rate of linear measure longimetry.
It should be noted that the loss function about various tasks, can use mask-rcnn, deeplab, Human
Parsing scheduling algorithm is constructed.
Step 633, the penalty values of the loss function are obtained.
For example, the parameter of random initializtion basic model calculates the damage of loss function in conjunction with first training sample
Mistake value.
If the penalty values of the loss function indicate the loss function convergence, i.e. the penalty values of loss function reach most
It is small, then it jumps and executes step 635.
Otherwise, if the penalty values of loss function indicate that the loss function is not converged, i.e., the penalty values of loss function are not
Reach minimum, then jumps and execute step 637.
Step 635, it is restrained to obtain described image parted pattern by the basic model.
Step 637, the parameter of the basic model is updated, and combines the latter training sample, continues to construct the basis
The loss function of model, until the penalty values of the loss function restrain.
It is noted that if the number of iterations has reached iteration threshold before the penalty values convergence of loss function,
Also stopping is continued to optimize to the parameter of basic model, guarantees the efficiency of model training with this.
So, it when loss function is restrained and meets required precision, indicates that model training is completed, thus obtains image
Parted pattern, so that the Image Segmentation Model is provided with the ability of area classification prediction.
In addition, in above process, based on the topological structure of entire neural network, so that image segmentation is end-to-end mistake
Journey, has fully ensured that the generalization and stability of image segmentation, and then is conducive to improve the accuracy rate of image segmentation.
Figure 13 is please referred to, in one exemplary embodiment, step 350 may comprise steps of:
Step 351, respectively to the target object area and it is described carry out curve fitting referring to object area, obtain correspond to institute
State the first fitted area of target object area and corresponding to second fitted area referring to object area.
It it is appreciated that either target object area is also with reference to object area, is indicated by pixel, and unstructured table
Show, is unfavorable for length acquisition.
For this purpose, using curve-fitting method, structured representation is carried out to the pixel in target object area in the present embodiment,
And structured representation is carried out to referring to the pixel in object area, to obtain the first fitted area and the second fitted area.
Optionally, curve-fitting method includes: least-square fitting approach, curve-fitting method based on Ransac etc..
Step 353, according to first fitted area and second fitted area, the picture of the target object area is obtained
Plain length and the length in pixels referring to object area.
It is considered as fitted area, target object area and ginseng in description, the first fitted area and the second fitted area for convenience
It is considered as candidate region according to object area.
Then, by curve matching, fitted area can be the external ellipse of minimum for being external in candidate region, be also possible to
It is external in the minimum circumscribed rectangle of candidate region.
Correspondingly, the length in pixels of candidate region can refer to the length of long side in minimum circumscribed rectangle, can also be
The length of long axis in minimum external ellipse, is also possible to the evolution of the product of long axis and short axle in minimum external ellipse.
Using object as pig, the length of object is that the long citing of pig body is illustrated, as shown in Figure 14 (a) to Figure 14 (b),
Candidate region forms a rectangle frame by curve matching, i.e. fitted area is to be external in the minimum circumscribed rectangle of candidate region,
At this point, in the minimum circumscribed rectangle long side length, be just considered as the length in pixels of candidate region.
Specifically, first fitted area is the first rectangle, and second fitted area is the second rectangle.If pig body
A length of pig's head is to the length of pigtail, then the first rectangle is 702, and the second rectangle is 701, as shown in Figure 14 (a)；If pig body is a length of
The pig basal part of the ear is to the length of pigtail, then the first rectangle is 703, and the second rectangle is 701, as shown in Figure 14 (b).
Using the length L1 of long side in first rectangle as the length in pixels of the target object area, and with described second
The length L2 of long side is as the length in pixels referring to object area in rectangle.
Under the action of above-described embodiment, the length in pixels, the length in pixels referring to object area that realize target object area
Calculating so that the calculating of the physical length of object is achieved.
Figure 15 to Figure 16 is a kind of specific implementation schematic diagram of length acquisition methods in an application scenarios.The application scenarios are
One is the settlement of insurance claim scene of livestock for object, wherein livestock is pig, pig body is long include the pig basal part of the ear to pigtail length,
Length of the pig's head to pigtail.
In settlement of insurance claim scene, including request end, Claims Resolution end and test side, wherein Claims Resolution end can be insurance company,
Test side is then the server for carrying out the long detection of pig body of insurance company's deployment, and it is pig throwing that request end, which then can be in insurance company,
The interaction timing of the user of guarantor, each section are as shown in figure 15.
Specifically, request end is after growing and fattening pigs are insured, if growing and fattening pigs are died of illness, request end can request Claims Resolution end needle
Settlement of insurance claim is carried out to the growing and fattening pigs died of illness, enters settlement of insurance claim link as a result,.
For request end, the image comprising object of reference and dead pig is acquired first, and insurance is initiated to Claims Resolution end with this
Claims Resolution request.
Claims Resolution end, then request in response to the settlement of insurance claim, the image of dead pig be sent to test side, is based on test side dead
The image of pig carries out the long automatic detection of pig body, specific implementation as in Figure 16 step 801 to shown in step 806.
After the estimation body that test side returns to dead pig is long, Claims Resolution end can be directed to damned pig based on long determine of the estimation body
Amount for which loss settled, and determining amount for which loss settled is transferred to request end.
As a result, it is that request end is completed for the settlement of insurance claim of growing and fattening pigs died of illness, in the settlement of insurance claim, supports to be directed to pig
The Claims Resolution mode that the basal part of the ear is detected to the length of pigtail and two boar body of the length length of pig's head to pigtail.
In this application scene, the long automatic detection of pig body is realized, can not only be avoided because water filling makes pig into pig body
The insurance fraud phenomenon that the human factors such as weight gain cause pig weight to lay particular stress on occurs, caused by being effectively reduced because of interference from human factor
Cost of settling a claim is excessively high, and avoids depending on artificial realization unduly in the long detection process of pig body, and then be effectively reduced settlement of insurance claim
Human cost in the process.
In addition, can not only accurately manage settlement of insurance claim link, and can ensure user for insurance company
The authenticity of Claims Resolution data, to effectively improve insurance company's high phenomenon of cost in settlement of insurance claim link.
Following is apparatus of the present invention embodiment, can be used for executing length acquisition methods according to the present invention.For this
Undisclosed details in invention device embodiment please refers to the embodiment of the method for length acquisition methods according to the present invention.
Figure 17 is please referred to, in one exemplary embodiment, a kind of length acquisition device 900 includes but is not limited to: image obtains
Modulus block 910, image segmentation module 930, length in pixels computing module 950 and object length computation module 970.
Wherein, image collection module 910, for obtaining the image comprising object and object of reference.
Image segmentation module 930, for be based on Image Segmentation Model, from described image segmentation obtain target object area and
Referring to object area.
Length in pixels computing module 950, for obtaining the reality of the length in pixels referring to object area and the object of reference
Border length, and obtain the length in pixels object of the target object area.
Object length computation module 970, in conjunction with the length in pixels referring to object area and the object of reference
Physical length calculates the physical length of the object according to the length in pixels of the target object area.
In one exemplary embodiment, device 900 as described above further includes but is not limited to: Zoom module and amendment mould
Block.
Wherein, Zoom module, the correction factor for being obtained based on study, contracts to the physical length of the object
Put processing.
Correction module, for scaling processing result to be updated to the physical length of the object.
In one exemplary embodiment, device 900 as described above further includes but is not limited to: learning sample obtain module and
Correction factor study module.
Wherein, learning sample obtains module, for obtaining sample to be learned.
Correction factor study module, for according to the physical length of object to be learned in the sample to be learned and true
Length carries out the study of the correction factor.
In one exemplary embodiment, the correction factor study module includes but is not limited to: relative error computing unit
And average calculation unit.
Wherein, relative error computing unit obtains in the sample to be learned for being directed to each sample to be learned wait learn
Practise the relative error between the physical length and actual length of object.
Average calculation unit, for carrying out processing of averaging to the relative error, using the average value as institute
State correction factor.
In one exemplary embodiment, described image segmentation module 930 includes but is not limited to: input unit, local feature
Extraction unit, global characteristics extraction unit and area classification predicting unit.
Wherein, input unit, for described image to be transmitted to convolutional layer by the input layer of described image parted pattern.
Local shape factor unit obtains the figure for carrying out feature extraction to described image by the convolutional layer
The local feature of picture.
Global characteristics extraction unit is obtained for carrying out Fusion Features by local feature of the articulamentum to described image
The global characteristics of described image.
Area classification predicting unit, for carrying out area classification prediction by global characteristics of the output layer to described image,
Obtain target object area in described image and referring to object area.
In one exemplary embodiment, device 900 as described above further includes but is not limited to: training sample obtain module and
Model training module.
Wherein, training sample obtains module, and for obtaining training sample, the training sample is to have carried out target object area
The image of mark and object of reference area marking.
Model training module carries out mould to the basic model for constructing basic model, and using the training sample
Type training obtains described image parted pattern.
In one exemplary embodiment, the model training module includes but is not limited to: loss function construction unit, loss
It is worth computing unit, model convergence unit and parameter updating unit.
Wherein, loss function construction unit, for based on the parameter for working as previous training sample and the basic model, structure
Build the loss function of the basic model.
Penalty values computing unit, for obtaining the penalty values of the loss function.
Model restrains unit, if the penalty values for the loss function restrain, is restrained by the basic model
To described image parted pattern.
Parameter updating unit updates institute if the penalty values for loss function indicate that the loss function is not converged
The parameter of basic model is stated, and combines the latter training sample, continues the loss function for constructing the basic model.
In one exemplary embodiment, the loss function of the basic model is the Classification Loss predicted for area classification
Function, the loss function that separating mask is introduced to the loss function of candidate region progress position recurrence, on the candidate region
The sum of, wherein the candidate region includes the target object area, the reference object area.
In one exemplary embodiment, the length in pixels computing module includes but is not limited to: curve matching unit and picture
Plain length acquiring unit.
Wherein, curve matching unit, for quasi- to the target object area and the reference object area march line respectively
It closes, obtains corresponding to the first fitted area of the target object area and corresponding to the second fitting area referring to object area
Domain.
Length in pixels acquiring unit, it is described for obtaining according to first fitted area and second fitted area
The length in pixels of target object area and the length in pixels referring to object area.
In one exemplary embodiment, first fitted area is the first rectangle, and second fitted area is second
Rectangle.
Correspondingly, the length in pixels acquiring unit includes but is not limited to: length defines subelement.
Wherein, length defines subelement, for using the length of long side in first rectangle as the target object area
Length in pixels, and using the length of long side in second rectangle as the length in pixels referring to object area.
In one exemplary embodiment, the object length computation module 970 includes but is not limited to: multiplying unit
With division arithmetic unit.
Wherein, multiplying unit, the actual (tube) length for length in pixels and the object of reference to the target object area
Degree carries out multiplying.
Division arithmetic unit, for carrying out division fortune to multiplication result and the length in pixels referring to object area
It calculates, using division arithmetic result as the physical length of the object.
In one exemplary embodiment, the object is pig, and the length of the object is that pig body is long, and the pig body is long
The length of length, pig's head including the pig basal part of the ear to pigtail to pigtail.
It should be noted that length acquisition device provided by above-described embodiment is when detecting object length, only more than
The division progress of each functional module is stated for example, can according to need and in practical application by above-mentioned function distribution by difference
Functional module complete, i.e., the internal structure of length acquisition device will be divided into different functional modules, to complete above description
All or part of function.
In addition, the embodiment of object length acquisition device provided by above-described embodiment and object length acquisition methods
Belonging to same design, the concrete mode that wherein modules execute operation is described in detail in embodiment of the method,
Details are not described herein again.
The embodiment of the invention also provides a kind of settlement of insurance claim system, settlement of insurance claim system include request end, Claims Resolution end and
Test side, wherein
Request end, for initiating settlement of insurance claim request to Claims Resolution end, carrying in settlement of insurance claim request includes livestock and reference
The image of object；
Image is sent to the test side for requesting in response to settlement of insurance claim by Claims Resolution end；
Test side, for carrying out the long detection of livestock body according to image, the long detection of livestock body includes: based on image segmentation mould
Type, segmentation obtains livestock region and referring to object area, obtains target object area by livestock region disconnecting, and obtain mesh from image
The length in pixels of object area and the length referring to object area are marked, in conjunction with the length in pixels referring to object area and the object of reference
Physical length, the estimation body for obtaining the livestock according to the length in pixels of the target object area is long；
Claims Resolution end, the estimation body for being also used to receive the livestock of test side return is long, and carries out according to the estimation body of livestock is long
Settlement of insurance claim.Specifically, Claims Resolution end calculates amount for which loss settled according to the estimation body of the livestock is long, the amount for which loss settled is transferred to
The request end.
Figure 18 is please referred to, in one exemplary embodiment, a kind of electronic equipment 1000, including but not limited to: at least one
Manage device 1001, at least a memory 1002 and an at least communication bus 1003.
Wherein, computer program is stored on memory 1002, processor 1001 reads storage by communication bus 1003
The computer program stored in device 1002.
The length acquisition methods in the various embodiments described above are realized when the computer program is executed by processor 1001.
In one exemplary embodiment, a kind of storage medium, is stored thereon with computer program, which is located
Manage the length acquisition methods realized in the various embodiments described above when device executes.
Above content, preferable examples embodiment only of the invention, is not intended to limit embodiment of the present invention, this
Field those of ordinary skill central scope according to the present invention and spirit can be carried out very easily corresponding flexible or repaired
Change, therefore protection scope of the present invention should be subject to protection scope required by claims.
Claims (15)
1. a kind of length acquisition methods characterized by comprising
Obtain the image comprising object and object of reference；
Based on Image Segmentation Model, segmentation obtains target object area and referring to object area from described image；
The physical length of the length in pixels referring to object area and the object of reference is obtained, and obtains the target object area
Length in pixels；
In conjunction with the physical length of the length in pixels referring to object area and the object of reference, according to the picture of the target object area
Plain length obtains the physical length of the object.
2. the method as described in claim 1, which is characterized in that described in the combination referring to object area length in pixels with it is described
The physical length of object of reference, after the physical length that the object is obtained according to the length in pixels of the target object area, institute
State method further include:
Based on the correction factor that study obtains, processing is zoomed in and out to the physical length of the object；
Scaling processing result is updated to the physical length of the object.
3. method according to claim 2, which is characterized in that the correction factor obtained based on study, to the target
Before the physical length of object zooms in and out processing, the method also includes:
Obtain sample to be learned；
According to the physical length and actual length of object to be learned in the sample to be learned, the correction factor is carried out
It practises.
4. method as claimed in claim 3, which is characterized in that described according to object to be learned in the sample to be learned
Physical length and actual length carry out the study of the correction factor, comprising:
For each sample to be learned, obtain in the sample to be learned the physical length of object to be learned and actual length it
Between relative error；
Processing of averaging is carried out to the relative error, using the average value as the correction factor.
5. such as the described in any item methods of Claims 1-4, which is characterized in that it is described to be based on Image Segmentation Model, from the figure
Segmentation obtains target object area and referring to object area as in, comprising:
Described image is transmitted to convolutional layer by the input layer of described image parted pattern；
Feature extraction is carried out to described image by the convolutional layer, obtains the local feature of described image；
Fusion Features are carried out by local feature of the articulamentum to described image, obtain the global characteristics of described image；
Area classification prediction is carried out by global characteristics of the output layer to described image, obtains the target object area in described image
With reference object area.
6. such as the described in any item methods of Claims 1-4, which is characterized in that the method also includes:
Training sample is obtained, the training sample is the image for having carried out target object area mark and object of reference area marking；
Basic model is constructed, and model training is carried out to the basic model using the training sample, obtains described image point
Cut model.
7. method as claimed in claim 6, which is characterized in that described to be carried out using the training sample to the basic model
Model training obtains described image parted pattern, comprising:
Based on the parameter for working as previous training sample and the basic model, the loss function of the basic model is constructed；
Obtain the penalty values of the loss function；
If the penalty values of the loss function indicate the loss function convergence, restrain to obtain by the basic model described
Image Segmentation Model；
Otherwise, the parameter of the basic model is updated, and combines the latter training sample, continues the damage for constructing the basic model
Lose function.
8. the method for claim 7, which is characterized in that the loss function of the basic model is pre- for area classification
The Classification Loss function of survey carries out the loss function of position recurrence to candidate region, introduces to separate on the candidate region and cover
The sum of loss function of mould, wherein the candidate region includes the target object area, the reference object area.
9. such as the described in any item methods of Claims 1-4, which is characterized in that the pixel for obtaining the target object area
Length and the length in pixels referring to object area, comprising:
Respectively to the target object area and it is described carry out curve fitting referring to object area, obtain correspond to the target object area
The first fitted area and corresponding to second fitted area referring to object area；
According to first fitted area and second fitted area, the length in pixels of the target object area and described is obtained
Referring to the length in pixels of object area.
10. method as claimed in claim 9, which is characterized in that first fitted area is the first rectangle, and described second is quasi-
Conjunction region is the second rectangle；
It is described according to first fitted area and second fitted area, obtain the target object area length in pixels and
The length in pixels referring to object area, comprising:
Using the length of long side in first rectangle as the length in pixels of the target object area, and in second rectangle
The length of long side is as the length in pixels referring to object area.
11. such as the described in any item methods of Claims 1-4, which is characterized in that referring to the pixel of object area described in the combination
The physical length of length and the object of reference obtains the actual (tube) length of the object according to the length in pixels of the target object area
Degree, comprising:
The physical length of length in pixels and the object of reference to the target object area carries out multiplying；
Division arithmetic is carried out to multiplication result and the length in pixels referring to object area, using division arithmetic result as institute
State the physical length of object.
12. a kind of length acquisition device characterized by comprising
Image collection module, for obtaining the image comprising object and object of reference；
Image segmentation module, for being based on Image Segmentation Model, segmentation obtains target object area and object of reference from described image
Region；
Length in pixels computing module, for obtaining the physical length of the length in pixels referring to object area and the object of reference,
And obtain the length in pixels of the target object area；
Object length computation module, for the actual (tube) length in conjunction with the length in pixels referring to object area and the object of reference
Degree, the physical length of the object is obtained according to the length in pixels of the target object area.
13. a kind of settlement of insurance claim system, which is characterized in that the settlement of insurance claim system includes request end, Claims Resolution end and test side,
Wherein,
The request end, for initiating settlement of insurance claim request to the Claims Resolution end, carrying in the settlement of insurance claim request includes domestic animal
The image of poultry and object of reference；
Described image is sent to the test side for requesting in response to the settlement of insurance claim by the Claims Resolution end；
The test side, for carrying out the long detection of livestock body according to described image, the long detection of the livestock body includes: based on image
Parted pattern, segmentation obtains livestock region and referring to object area, obtains target by the livestock region disconnecting from described image
Object area, and the length in pixels and the length in pixels referring to object area of the target object area are obtained, in conjunction with the reference
The physical length of the length in pixels of object area and the object of reference obtains the domestic animal according to the length in pixels of the target object area
The estimation body of poultry is long；
The Claims Resolution end, the estimation body for being also used to receive the livestock that the test side returns is long, and according to the livestock
Estimate the long progress settlement of insurance claim of body.
14. a kind of storage medium, which is characterized in that be stored thereon with computer program, the computer program is held by processor
Length acquisition methods as described in any one of claim 1 to 11 are realized when row.
15. a kind of electronic equipment characterized by comprising
Processor；
Memory is stored with computer program on the memory, and the computer program is realized when being executed by the processor
Length acquisition methods as described in any one of claim 1 to 11.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910271246.5A CN110175503A (en) | 2019-04-04 | 2019-04-04 | Length acquisition methods, device, settlement of insurance claim system, medium and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910271246.5A CN110175503A (en) | 2019-04-04 | 2019-04-04 | Length acquisition methods, device, settlement of insurance claim system, medium and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110175503A true CN110175503A (en) | 2019-08-27 |
Family
ID=67689416
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910271246.5A Pending CN110175503A (en) | 2019-04-04 | 2019-04-04 | Length acquisition methods, device, settlement of insurance claim system, medium and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110175503A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111198549A (en) * | 2020-02-18 | 2020-05-26 | 陈文翔 | Poultry breeding monitoring management system based on big data |
| CN111369372A (en) * | 2020-03-03 | 2020-07-03 | 福州农福腾信息科技有限公司 | Insurance claim settlement method based on pig body recognition and background terminal |
| CN111860652A (en) * | 2020-07-22 | 2020-10-30 | 中国平安财产保险股份有限公司 | Method, device, equipment and medium for measuring animal body weight based on image detection |
| CN114926633A (en) * | 2022-03-25 | 2022-08-19 | 成都爱记科技有限公司 | High-precision pig weight estimation method based on deep learning |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2012067374A2 (en) * | 2010-11-16 | 2012-05-24 | Lg Innotek Co., Ltd. | Camera module and method for correcting lens shading thereof |
| US20120275659A1 (en) * | 2011-04-27 | 2012-11-01 | Steve Gomas | Apparatus and method for estimation of livestock weight |
| US20140193039A1 (en) * | 2013-01-07 | 2014-07-10 | Ronald M. Wexler | System and method of measuring distances related to an object |
| CN106885515A (en) * | 2017-03-01 | 2017-06-23 | 苏州光照精密仪器有限公司 | A kind of image measuring method, device and system |
| CN108961330A (en) * | 2018-06-22 | 2018-12-07 | 深源恒际科技有限公司 | The long measuring method of pig body and system based on image |
| CN109165645A (en) * | 2018-08-01 | 2019-01-08 | 腾讯科技（深圳）有限公司 | A kind of image processing method, device and relevant device |
| US20190057507A1 (en) * | 2017-08-18 | 2019-02-21 | Samsung Electronics Co., Ltd. | System and method for semantic segmentation of images |
| CN109559342A (en) * | 2018-03-05 | 2019-04-02 | 北京佳格天地科技有限公司 | The long measurement method of animal body and device |
-
2019
- 2019-04-04 CN CN201910271246.5A patent/CN110175503A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2012067374A2 (en) * | 2010-11-16 | 2012-05-24 | Lg Innotek Co., Ltd. | Camera module and method for correcting lens shading thereof |
| US20120275659A1 (en) * | 2011-04-27 | 2012-11-01 | Steve Gomas | Apparatus and method for estimation of livestock weight |
| US20140193039A1 (en) * | 2013-01-07 | 2014-07-10 | Ronald M. Wexler | System and method of measuring distances related to an object |
| CN106885515A (en) * | 2017-03-01 | 2017-06-23 | 苏州光照精密仪器有限公司 | A kind of image measuring method, device and system |
| US20190057507A1 (en) * | 2017-08-18 | 2019-02-21 | Samsung Electronics Co., Ltd. | System and method for semantic segmentation of images |
| CN109559342A (en) * | 2018-03-05 | 2019-04-02 | 北京佳格天地科技有限公司 | The long measurement method of animal body and device |
| CN108961330A (en) * | 2018-06-22 | 2018-12-07 | 深源恒际科技有限公司 | The long measuring method of pig body and system based on image |
| CN109165645A (en) * | 2018-08-01 | 2019-01-08 | 腾讯科技（深圳）有限公司 | A kind of image processing method, device and relevant device |
Non-Patent Citations (3)
| Title |
|---|
| KYUNGKOO JUN 等: "Estimating pig weights from images without constraint on posture and illumination", 《ELSEVIER》 * |
| 司永胜等: "基于Kinect相机的猪体理想姿态检测与体尺测量", 《农业机械学报》, vol. 50, no. 1 * |
| 王福昌 等主编: "《智能图像处理及应用》", 华南理工大学出版社, pages: 178 - 181 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111198549A (en) * | 2020-02-18 | 2020-05-26 | 陈文翔 | Poultry breeding monitoring management system based on big data |
| CN111198549B (en) * | 2020-02-18 | 2020-11-06 | 湖南伟业动物营养集团股份有限公司 | Poultry breeding monitoring management system based on big data |
| CN111369372A (en) * | 2020-03-03 | 2020-07-03 | 福州农福腾信息科技有限公司 | Insurance claim settlement method based on pig body recognition and background terminal |
| CN111369372B (en) * | 2020-03-03 | 2023-02-07 | 福州农福腾信息科技有限公司 | Insurance claim settlement method based on pig body recognition and background terminal |
| CN111860652A (en) * | 2020-07-22 | 2020-10-30 | 中国平安财产保险股份有限公司 | Method, device, equipment and medium for measuring animal body weight based on image detection |
| CN111860652B (en) * | 2020-07-22 | 2022-03-29 | 中国平安财产保险股份有限公司 | Method, device, equipment and medium for measuring animal body weight based on image detection |
| CN114926633A (en) * | 2022-03-25 | 2022-08-19 | 成都爱记科技有限公司 | High-precision pig weight estimation method based on deep learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110175503A (en) | Length acquisition methods, device, settlement of insurance claim system, medium and electronic equipment | |
| CN108537794B (en) | Medical image data processing method, apparatus and computer readable storage medium | |
| CN111127527B (en) | Method and device for realizing lung nodule self-adaptive matching based on CT image bone registration | |
| CN109141248A (en) | Pig weight measuring method and system based on image | |
| CN108871520A (en) | Livestock body weight measurement and device | |
| CN109583594B (en) | Deep learning training method, device, equipment and readable storage medium | |
| CN111325217B (en) | Data processing method, device, system and medium | |
| CN111539937A (en) | Object index detection method and livestock weight detection method and device | |
| CN114446440A (en) | Radiotherapy plan determination method and device, storage medium and processor | |
| CN109272546A (en) | A kind of fry length measurement method and system | |
| CN111340209A (en) | Network model training method, image segmentation method and focus positioning method | |
| CN112233149A (en) | Scene flow determination method and device, storage medium and electronic device | |
| KR102297317B1 (en) | Livestock weighing device and method through 3D image processing | |
| CN109408041A (en) | A kind of the map coordinates system conversion method and electronic equipment of language based on programming | |
| CN109657535A (en) | Image identification method, target device and cloud platform | |
| CN112261399A (en) | Capsule endoscope image three-dimensional reconstruction method, electronic device and readable storage medium | |
| CN116649953A (en) | Wound scanning method and device and wound scanner | |
| CN112461342A (en) | Aquatic product weighing method, terminal equipment and storage medium | |
| Zhang et al. | Body weight estimation of yak based on cloud edge computing | |
| KR102597931B1 (en) | Method for image analyzing about luxating patella | |
| CN111833395B (en) | Direction-finding system single target positioning method and device based on neural network model | |
| CN115081621A (en) | Model training method, focus segmentation device, computer device, and medium | |
| CN107194931A (en) | It is a kind of that the method and system for obtaining target depth information is matched based on binocular image | |
| Minghuang et al. | Automatic Measurement of Mongolian Horse Body Based on Improved YOLOv8n-pose and Point Cloud Analysis | |
| CN111507432A (en) | Intelligent weighing method and system for agricultural insurance claims, electronic equipment and 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 |