CN109635813A - A kind of steel rail area image partition method and device - Google Patents
A kind of steel rail area image partition method and device Download PDFInfo
- Publication number
- CN109635813A CN109635813A CN201811523003.8A CN201811523003A CN109635813A CN 109635813 A CN109635813 A CN 109635813A CN 201811523003 A CN201811523003 A CN 201811523003A CN 109635813 A CN109635813 A CN 109635813A
- Authority
- CN
- China
- Prior art keywords
- rail
- autocoder
- image
- sample image
- test
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The present invention discloses a kind of steel rail area image partition method and device, and method includes: first autocoder of input selected in rail sample image, for express sample image pattern feature and export selection rail sample image steel rail area segmentation figure；It obtains test rail image and inputs the second autocoder, for expressing test image pattern feature and exporting the segmentation figure of test rail image steel rail area；The rail sample image of selection and test rail image are compared using encoder is compared, the segmentation figure of steel rail area in two images is obtained, wherein the input for comparing each characteristic layer of encoder is to incorporate the first autocoder for having inputted the rail sample image and the output of the character pair layer of the second autocoder of input test rail image according to default amalgamation mode；Compare coder-decoder part and the first and second equal shared parameter of autocoder decoded portion.It is able to achieve the extraction of high-precision, the steel rail area of strong robust in picture.
Description
Technical field
The present embodiments relate to technical field of computer vision, and in particular to a kind of steel rail area image partition method and
Device.
Background technique
In recent years, rail traffic because its have large conveying quantity and fast and safely many advantages, such as obtained fast violent development, give
The production and living of people bring great convenience.
In field of track traffic, Analysis of Rail Surface Abrasion is of great significance, thus rail table is partitioned into picture
Face region has significant application value.
In consideration of it, how to realize high-precision in picture, the extraction of the steel rail area of strong robust becomes what needs at present solved
Technical problem.
Summary of the invention
Since existing method is there are the above problem, the embodiment of the present invention proposes a kind of steel rail area image partition method and dress
It sets.
In a first aspect, the embodiment of the present invention proposes a kind of steel rail area image partition method, comprising:
One in rail sample image is selected, selected rail sample image is inputted into the first autocoder, institute
The first autocoder is stated for expressing the pattern feature of sample image, and exports the rail area of selected rail sample image
The segmentation figure in domain；
A test rail image is obtained, the test rail image is inputted into the second autocoder, described second certainly
Dynamic encoder is used to express the pattern feature of test image, and exports the segmentation figure of the steel rail area of the test rail image；
Selected rail sample image and an acquired test rail image are compared using encoder is compared
It is right, obtain the segmentation figure of steel rail area in two images, wherein the input for comparing each characteristic layer of encoder, is according to pre-
If amalgamation mode incorporate the first autocoder for having inputted the rail sample image and inputted the test rail
The output of the character pair layer of second autocoder of image；
Wherein, the decoder section for comparing encoder and first autocoder and the second autocoder
The equal shared parameter of decoded portion.
Second aspect, the embodiment of the present invention also propose a kind of steel rail area image segmentation device, comprising:
First input module inputs selected rail sample image for selecting one in rail sample image
First autocoder, first autocoder are used to express the pattern feature of sample image, and export selected steel
The segmentation figure of the steel rail area of rail sample image；
Second input module, it is for obtaining a test rail image, test rail image input second is automatic
Encoder, second autocoder are used to express the pattern feature of test image, and export the test rail image
The segmentation figure of steel rail area；
Output module compares encoder for selected rail sample image and an acquired tested steel for utilizing
Rail image is compared, and obtains the segmentation figure of steel rail area in two images, wherein described to compare the defeated of each characteristic layer of encoder
Enter, is to incorporate the first autocoder for having inputted the rail sample image according to preset amalgamation mode and inputted institute
State the output of the character pair layer of the second autocoder of test rail image；
Wherein, the decoder section for comparing encoder and first autocoder and the second autocoder
The equal shared parameter of decoded portion.
The third aspect, the embodiment of the present invention provide a kind of electronic equipment, comprising: processor, memory, bus and are stored in
On memory and the computer program that can be run on processor；
Wherein, the processor, memory complete mutual communication by the bus；
The processor realizes the above method when executing the computer program.
Fourth aspect, the embodiment of the present invention provide a kind of non-transient computer readable storage medium, on the storage medium
It is stored with computer program, which realizes the above method when being executed by processor.
As shown from the above technical solution, a kind of steel rail area image partition method and device provided in an embodiment of the present invention,
By one in selection rail sample image, selected rail sample image is inputted into the first autocoder, described the
One autocoder is used to express the pattern feature of sample image and exports the steel rail area of selected rail sample image
Segmentation figure obtains a test rail image, the test rail image is inputted the second autocoder, described second is automatic
Encoder is used to express the pattern feature of test image and exports the segmentation figure of the steel rail area of the test rail image, utilizes
It compares encoder selected rail sample image is compared with an acquired test rail image, obtains two figures
The segmentation figure of steel rail area as in, wherein the input for comparing each characteristic layer of encoder is incorporated according to default amalgamation mode
It inputs the first autocoder of selected rail sample image and has inputted the of an acquired test rail image
The output of the character pair layer of two autocoders；The decoder section for comparing encoder and first autocoder
With the equal shared parameter of decoded portion of the second autocoder, thereby, it is possible to realize the rail area of high-precision in picture, strong robust
The extraction in domain, speed is fast, adaptable.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with
Other attached drawings are obtained according to these figures.
Fig. 1 is a kind of flow diagram for steel rail area image partition method that one embodiment of the invention provides；
Fig. 2 is the schematic illustration provided in an embodiment of the present invention for comparing encoder；
Fig. 3 is a kind of structural schematic diagram for steel rail area image segmentation device that one embodiment of the invention provides；
Fig. 4 is the entity structure schematic diagram for the electronic equipment that one embodiment of the invention provides.
Specific embodiment
With reference to the accompanying drawing, further description of the specific embodiments of the present invention.Following embodiment is only used for more
Technical solution of the present invention is clearly demonstrated, and not intended to limit the protection scope of the present invention.
Fig. 1 shows a kind of flow diagram of steel rail area image partition method of one embodiment of the invention offer, such as
Shown in Fig. 1, the steel rail area image partition method of the present embodiment, comprising:
One in S1, selection rail sample image, selected rail sample image is inputted into the first autocoder,
First autocoder is used to express the pattern feature of sample image, and exports the rail of selected rail sample image
The segmentation figure in region.
S2, obtain a test rail image, by the test rail image input the second autocoder, described second
Autocoder is used to express the pattern feature of test image, and exports the segmentation of the steel rail area of the test rail image
Figure.
S3, selected rail sample image and an acquired test rail image are carried out using comparison encoder
Compare, obtain the segmentation figure of steel rail area in two images, wherein it is described compare each characteristic layer of encoder input, be according to
Preset amalgamation mode incorporates the first autocoder for having inputted the rail sample image and has inputted the tested steel
The output of the character pair layer of second autocoder of rail image；
Wherein, the decoder section for comparing encoder and first autocoder and the second autocoder
The equal shared parameter of decoded portion.
In a particular application, first autocoder and second autocoder are all based on sample image pair
What deep neural network obtained after being trained；The sample image is by obtaining rail sample image, to rail sample graph
As the segmentation figure for being labeled with steel rail area obtained after artificial mark segmentation.
It is understood that be all perpendicular strip region since steel rail area image on piece has the similar general character of structure,
Therefore the present embodiment can by allowing comparison encoder association contrast images content in advance, find out similar perpendicular strip region from
And realize the extraction of steel rail area.
Steel rail area image partition method provided in an embodiment of the present invention, by selecting one in rail sample image,
Selected rail sample image is inputted into the first autocoder, first autocoder is for expressing sample image
Pattern feature and export selected rail sample image steel rail area segmentation figure, obtain a test rail image, will
The test rail image inputs the second autocoder, and the mode that second autocoder is used to express test image is special
The segmentation figure for levying and exporting the steel rail area of the test rail image, using comparing encoder for selected rail sample graph
As being compared with an acquired test rail image, the segmentation figure of steel rail area in two images is obtained, wherein compare
The input of each characteristic layer of encoder is that inputted selected rail sample image first is incorporated according to default amalgamation mode
Autocoder and inputted an acquired test rail image the second autocoder character pair layer output；
The decoder section for comparing encoder and the decoded portion of first autocoder and the second autocoder are total
Parameter is enjoyed, thereby, it is possible to realize the extraction of high-precision, the steel rail area of strong robust in picture, speed is fast, adaptable.
Further, in a particular application, it can refer to Fig. 2, it is described to have inputted the first automatic of the rail sample image
Encoder and inputted it is described test rail image the second autocoder character pair layer output, compare volume with described
The output of code device individual features layer forms the input for comparing the lower characteristic layer of encoder in such a way that channel is in parallel.
Further, in a particular application, the default amalgamation mode can be with are as follows: (compare the output-the first of encoder from
The output of dynamic encoder) ++ the output of the second autocoder, wherein ++ it is or operates.
The comparison encoder can learn contrast images content as a result, find out similar perpendicular strip region to realize
The extraction of steel rail area.
Further, on the basis of the above embodiments, before the step S1, the present embodiment the method can be with
Include the steps that S0 is not shown in the figure:
S0, acquisition sample image are trained deep neural network according to the sample image, obtain trained
The decoder section of first autocoder, first autocoder exports a channel, is responsible for the segmentation of steel rail area.
The present embodiment can obtain trained first autocoder as a result,.
Further, on the basis of the above embodiments, before the step S2, the present embodiment the method can be with
Include the steps that P1 is not shown in the figure:
It obtains sample image to be trained deep neural network according to the sample image, obtains trained second
The decoder section of autocoder, second autocoder exports a channel, is responsible for the segmentation of steel rail area；
Wherein, the parameter of second autocoder and first autocoder is shared.
It is understood that the realization side of the first autocoder of the embodiment of the present invention and second autocoder
Formula is not unique, various deep neural networks can be used usually to realize, for example, the deep neural network in the present embodiment
It may include: u-net, deconvolution neural network etc..
The present embodiment can obtain trained second autocoder as a result,.
Steel rail area image partition method provided in an embodiment of the present invention, takes full advantage of the texture paging of Rail Surface
With the consistency feature of whole rail form, encoder association contrast images content is compared by allowing in advance, can be realized picture
The extraction of middle high-precision, the steel rail area of strong robust, speed is fast, adaptable.
Fig. 3 shows a kind of structural schematic diagram of steel rail area image segmentation device of one embodiment of the invention offer, such as
Shown in Fig. 3, the steel rail area image segmentation device of the present embodiment, comprising: the first input module 31, the second input module 32 and defeated
Module 33 out；Wherein:
First input module 31, for selecting one in rail sample image, by selected rail sample image
The first autocoder is inputted, first autocoder is used to express the pattern feature of sample image, and selected by output
Rail sample image steel rail area segmentation figure；
The test rail image is inputted for obtaining a test rail image by second input module 32
Two autocoders, second autocoder are used to express the pattern feature of test image, and export the test rail
The segmentation figure of the steel rail area of image；
The output module 33 compares encoder for selected rail sample image and acquired one for utilizing
Test rail image is compared, and obtains the segmentation figure of steel rail area in two images, wherein each feature of the comparison encoder
The input of layer is that the first autocoder for having inputted the rail sample image and is incorporated according to preset amalgamation mode
Input the output of the character pair layer of the second autocoder of the test rail image；
Wherein, the decoder section for comparing encoder and first autocoder and the second autocoder
The equal shared parameter of decoded portion.
Specifically, first input module 31 selects one in rail sample image, by selected rail sample graph
As the first autocoder of input, first autocoder is used to express the pattern feature of sample image, and selected by output
The segmentation figure of the steel rail area for the rail sample image selected；Second input module 32 obtains a test rail image, will
The test rail image inputs the second autocoder, and the mode that second autocoder is used to express test image is special
Sign, and export the segmentation figure of the steel rail area of the test rail image；The output module 33, which utilizes, compares encoder for institute
The rail sample image of selection is compared with an acquired test rail image, obtains steel rail area in two images
Segmentation figure, wherein the input for comparing each characteristic layer of encoder, be incorporated according to preset amalgamation mode inputted it is described
First autocoder of rail sample image and the corresponding spy for having inputted second autocoder for testing rail image
Levy the output of layer；Wherein, the decoder section and first autocoder and the second autocoding for comparing encoder
The equal shared parameter of the decoded portion of device.
In a particular application, first autocoder and second autocoder are all based on sample image pair
What deep neural network obtained after being trained；The sample image is by obtaining rail sample image, to rail sample graph
As the segmentation figure for being labeled with steel rail area obtained after artificial mark segmentation.
It is understood that be all perpendicular strip region since steel rail area image on piece has the similar general character of structure,
Therefore the present embodiment can by allowing comparison encoder association contrast images content in advance, find out similar perpendicular strip region from
And realize the extraction of steel rail area.
Steel rail area image segmentation device provided in an embodiment of the present invention can be realized in picture high-precision, strong robust
The extraction of steel rail area, speed is fast, adaptable.
Further, in a particular application, it can refer to Fig. 2, it is described to have inputted the first automatic of the rail sample image
Encoder and inputted it is described test rail image the second autocoder character pair layer output, compare volume with described
The output of code device individual features layer forms the input for comparing the lower characteristic layer of encoder in such a way that channel is in parallel.
Further, in a particular application, the default amalgamation mode can be with are as follows: (compare the output-the first of encoder from
The output of dynamic encoder) ++ the output of the second autocoder, wherein ++ it is or operates.
The comparison encoder can learn contrast images content as a result, find out similar perpendicular strip region to realize
The extraction of steel rail area.
Further, on the basis of the above embodiments, the present embodiment described device can also include not shown in the figure:
First training module, according to the sample image, instructs deep neural network for obtaining sample image
Practice, obtain trained first autocoder, the decoder section of first autocoder exports a channel, is responsible for
The segmentation of steel rail area.
The present embodiment can obtain trained first autocoder as a result,.
Further, on the basis of the above embodiments, the present embodiment described device can also include not shown in the figure:
Second training module, according to the sample image, instructs deep neural network for obtaining sample image
Practice, obtain trained second autocoder, the decoder section of second autocoder exports a channel, is responsible for
The segmentation of steel rail area；
Wherein, the parameter of second autocoder and first autocoder is shared.
It is understood that the realization side of the first autocoder of the embodiment of the present invention and second autocoder
Formula is not unique, various deep neural networks can be used usually to realize, for example, the deep neural network in the present embodiment
It may include: u-net, deconvolution neural network etc..
The present embodiment can obtain trained second autocoder as a result,.
Steel rail area image segmentation device provided in an embodiment of the present invention, takes full advantage of the texture paging of Rail Surface
With the consistency feature of whole rail form, encoder association contrast images content is compared by allowing in advance, can be realized picture
The extraction of middle high-precision, the steel rail area of strong robust, speed is fast, adaptable.
The steel rail area image segmentation device of the present embodiment, can be used for executing the technical solution of preceding method embodiment,
That the realization principle and technical effect are similar is similar for it, and details are not described herein again.
Fig. 4 shows the entity structure schematic diagram of a kind of electronic equipment of one embodiment of the invention offer, as shown in figure 4,
The electronic equipment may include: processor 401, memory 402, bus 403 and be stored on memory 402 and can be in processor
The computer program run on 401；
Wherein, the processor 401 and memory 402 complete mutual communication by the bus 403；
The processor 401 realizes method provided by above method embodiment when executing the computer program, such as
Include: one selected in rail sample image, selected rail sample image is inputted into the first autocoder, described the
One autocoder is used to express the pattern feature of sample image, and export the steel rail area of selected rail sample image
Segmentation figure；A test rail image is obtained, the test rail image is inputted into the second autocoder, described second is automatic
Encoder is used to express the pattern feature of test image, and exports the segmentation figure of the steel rail area of the test rail image；Benefit
Selected rail sample image is compared with an acquired test rail image with encoder is compared, obtains two
The segmentation figure of steel rail area in image, wherein the input for comparing each characteristic layer of encoder, is according to preset amalgamation mode
Incorporate the first autocoder for having inputted the rail sample image and inputted it is described test rail image second from
The output of the character pair layer of dynamic encoder；Wherein, the decoder section and first autocoding for comparing encoder
The equal shared parameter of the decoded portion of device and the second autocoder.
The embodiment of the present invention provides a kind of non-transient computer readable storage medium, is stored thereon with computer program, should
Method provided by above method embodiment is realized when computer program is executed by processor, for example, selection rail sample
One in image, selected rail sample image is inputted into the first autocoder, first autocoder is used for
The pattern feature of sample image is expressed, and exports the segmentation figure of the steel rail area of selected rail sample image；Obtain one
Rail image is tested, the test rail image is inputted into the second autocoder, second autocoder is for expressing
The pattern feature of test image, and export the segmentation figure of the steel rail area of the test rail image；It will using encoder is compared
Selected rail sample image is compared with an acquired test rail image, obtains steel rail area in two images
Segmentation figure, wherein it is described compare each characteristic layer of encoder input, be to incorporate to have inputted institute according to preset amalgamation mode
It states the first autocoder of rail sample image and has inputted the correspondence of the second autocoder of the test rail image
The output of characteristic layer；Wherein, the decoder section for comparing encoder and first autocoder and the second automatic volume
The equal shared parameter of decoded portion of code device.
It should be understood by those skilled in the art that, embodiments herein can provide as method, apparatus or computer program
Product.Therefore, complete hardware embodiment, complete software embodiment or reality combining software and hardware aspects can be used in the application
Apply the form of example.Moreover, it wherein includes the computer of computer usable program code that the application, which can be used in one or more,
The computer program implemented in usable storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) produces
The form of product.
The application be referring to according to the method, apparatus of the embodiment of the present application and the flow chart of computer program product and/or
Block diagram describes.It should be understood that each process that can be realized by computer program instructions in flowchart and/or the block diagram and/or
The combination of process and/or box in box and flowchart and/or the block diagram.It can provide these computer program instructions to arrive
General purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices processor to generate one
Machine, so that being generated by the instruction that computer or the processor of other programmable data processing devices execute for realizing flowing
The device/system for the function of being specified in journey figure one process or multiple processes and/or block diagrams one box or multiple boxes.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or
The instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram one
The step of function of being specified in a box or multiple boxes.
It should be noted that, in this document, relational terms such as first and second and the like are used merely to a reality
Body or operation are distinguished with another entity or operation, are deposited without necessarily requiring or implying between these entities or operation
In any actual relationship or order or sequence.Moreover, the terms "include", "comprise" or its any other variant are intended to
Non-exclusive inclusion, so that the process, method, article or equipment including a series of elements is not only wanted including those
Element, but also including other elements that are not explicitly listed, or further include for this process, method, article or equipment
Intrinsic element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that
There is also other identical elements in process, method, article or equipment including the element.Term " on ", "lower" etc. refer to
The orientation or positional relationship shown is to be based on the orientation or positional relationship shown in the drawings, and is merely for convenience of the description present invention and simplifies
Description, rather than the device or element of indication or suggestion meaning must have a particular orientation, constructed and grasped with specific orientation
Make, therefore is not considered as limiting the invention.Unless otherwise clearly defined and limited, term " installation ", " connected ",
" connection " shall be understood in a broad sense, for example, it may be being fixedly connected, may be a detachable connection, or be integrally connected；It can be
Mechanical connection, is also possible to be electrically connected；It can be directly connected, two can also be can be indirectly connected through an intermediary
Connection inside element.For the ordinary skill in the art, above-mentioned term can be understood at this as the case may be
Concrete meaning in invention.
In specification of the invention, numerous specific details are set forth.Although it is understood that the embodiment of the present invention can
To practice without these specific details.In some instances, well known method, structure and skill is not been shown in detail
Art, so as not to obscure the understanding of this specification.Similarly, it should be understood that disclose in order to simplify the present invention and helps to understand respectively
One or more of a inventive aspect, in the above description of the exemplary embodiment of the present invention, each spy of the invention
Sign is grouped together into a single embodiment, figure, or description thereof sometimes.However, should not be by the method solution of the disclosure
Release is in reflect an intention that i.e. the claimed invention requires more than feature expressly recited in each claim
More features.More precisely, as the following claims reflect, inventive aspect is less than single reality disclosed above
Apply all features of example.Therefore, it then follows thus claims of specific embodiment are expressly incorporated in the specific embodiment,
It is wherein each that the claims themselves are regarded as separate embodiments of the invention.It should be noted that in the absence of conflict, this
The feature in embodiment and embodiment in application can be combined with each other.The invention is not limited to any single aspect,
It is not limited to any single embodiment, is also not limited to any combination and/or displacement of these aspects and/or embodiment.And
And can be used alone each aspect and/or embodiment of the invention or with other one or more aspects and/or its implementation
Example is used in combination.
Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention., rather than its limitations；To the greatest extent
Pipe present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: its according to
So be possible to modify the technical solutions described in the foregoing embodiments, or to some or all of the technical features into
Row equivalent replacement；And these are modified or replaceed, various embodiments of the present invention technology that it does not separate the essence of the corresponding technical solution
The range of scheme should all cover within the scope of the claims and the description of the invention.
Claims (10)
1. a kind of steel rail area image partition method characterized by comprising
One in rail sample image is selected, selected rail sample image is inputted into the first autocoder, described the
One autocoder is used to express the pattern feature of sample image, and export the steel rail area of selected rail sample image
Segmentation figure；
A test rail image is obtained, the test rail image is inputted into the second autocoder, the described second automatic volume
Code device is used to express the pattern feature of test image, and exports the segmentation figure of the steel rail area of the test rail image；
Selected rail sample image is compared with an acquired test rail image using encoder is compared, is obtained
Take the segmentation figure of steel rail area in two images, wherein the input for comparing each characteristic layer of encoder, is melted according to preset
Conjunction mode incorporates the first autocoder for having inputted the rail sample image and has inputted the test rail image
The output of the character pair layer of second autocoder；
Wherein, the decoding of the decoder section for comparing encoder and first autocoder and the second autocoder
Partially equal shared parameter.
2. the method according to claim 1, wherein first autocoder and second autocoding
Device, which is all based on, to be obtained after sample image is trained deep neural network；
The sample image is by obtaining rail sample image, to what is obtained after the artificial mark segmentation of rail sample image progress
It is labeled with the segmentation figure of steel rail area.
3. the method according to claim 1, wherein described inputted the first automatic of the rail sample image
Encoder and inputted it is described test rail image the second autocoder character pair layer output, compare volume with described
The output of code device individual features layer forms the input for comparing the lower characteristic layer of encoder in such a way that channel is in parallel.
4. the method according to claim 1, wherein the default amalgamation mode are as follows: (compare the defeated of encoder
The output of-the first autocoder out) ++ the output of the second autocoder, wherein ++ it is or operates.
5., will be selected the method according to claim 1, wherein one in selection rail sample image
Rail sample image input the first autocoder before, the method also includes:
It obtains sample image to be trained deep neural network according to the sample image, it is automatic to obtain trained first
The decoder section of encoder, first autocoder exports a channel, is responsible for the segmentation of steel rail area.
6. according to the method described in claim 5, it is characterized in that, obtaining a test rail image, by the tested steel
Rail image inputs before the second autocoder, the method also includes:
It obtains sample image to be trained deep neural network according to the sample image, it is automatic to obtain trained second
The decoder section of encoder, second autocoder exports a channel, is responsible for the segmentation of steel rail area；
Wherein, the parameter of second autocoder and first autocoder is shared.
7. a kind of steel rail area image segmentation device characterized by comprising
Selected rail sample image is inputted first for selecting one in rail sample image by the first input module
Autocoder, first autocoder are used to express the pattern feature of sample image, and export selected rail sample
The segmentation figure of the steel rail area of this image；
The test rail image is inputted the second autocoding for obtaining a test rail image by the second input module
Device, second autocoder are used to express the pattern feature of test image, and export the rail of the test rail image
The segmentation figure in region；
Output module compares encoder for selected rail sample image and an acquired test rail figure for utilizing
As being compared, the segmentation figure of steel rail area in two images is obtained, wherein the input for comparing each characteristic layer of encoder,
It is to incorporate the first autocoder for having inputted the rail sample image according to preset amalgamation mode and inputted described
Test the output of the character pair layer of the second autocoder of rail image；
Wherein, the decoding of the decoder section for comparing encoder and first autocoder and the second autocoder
Partially equal shared parameter.
8. device according to claim 7, which is characterized in that described to have inputted the first automatic of the rail sample image
Encoder and inputted it is described test rail image the second autocoder character pair layer output, compare volume with described
The output of code device individual features layer forms the input for comparing the lower characteristic layer of encoder in such a way that channel is in parallel.
9. a kind of electronic equipment characterized by comprising processor, memory, bus and storage are on a memory and can be the
The computer program run on processor；
Wherein, the processor, memory complete mutual communication by the bus；
The processor realizes such as method of any of claims 1-6 when executing the computer program.
10. a kind of non-transient computer readable storage medium, which is characterized in that in the non-transient computer readable storage medium
It is stored with computer program, such as side of any of claims 1-6 is realized when which is executed by processor
Method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811523003.8A CN109635813B (en) | 2018-12-13 | 2018-12-13 | Steel rail region image segmentation method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811523003.8A CN109635813B (en) | 2018-12-13 | 2018-12-13 | Steel rail region image segmentation method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109635813A true CN109635813A (en) | 2019-04-16 |
| CN109635813B CN109635813B (en) | 2020-12-25 |
Family
ID=66073390
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811523003.8A Active CN109635813B (en) | 2018-12-13 | 2018-12-13 | Steel rail region image segmentation method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109635813B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101778275A (en) * | 2009-01-09 | 2010-07-14 | 深圳市融创天下科技发展有限公司 | Image processing method of self-adaptive time domain and spatial domain resolution ratio frame |
| US8432410B1 (en) * | 2006-07-28 | 2013-04-30 | Nvidia Corporation | 3D graphics API extension for a shared exponent image format |
| US20170019684A1 (en) * | 2014-05-20 | 2017-01-19 | Here Global B.V. | Predictive value data set compression |
| CN107357899A (en) * | 2017-07-14 | 2017-11-17 | 吉林大学 | Based on the short text sentiment analysis method with product network depth autocoder |
| CN107481188A (en) * | 2017-06-23 | 2017-12-15 | 珠海经济特区远宏科技有限公司 | A kind of image super-resolution reconstructing method |
| CN107977598A (en) * | 2016-10-21 | 2018-05-01 | 三星电子株式会社 | Method and apparatus for identifying facial expression |
| CN108280452A (en) * | 2018-01-26 | 2018-07-13 | 深圳市唯特视科技有限公司 | A kind of image, semantic label correction method based on parallel network framework |
| CN108510560A (en) * | 2018-04-11 | 2018-09-07 | 腾讯科技（深圳）有限公司 | Image processing method, device, storage medium and computer equipment |
| CN108830221A (en) * | 2018-06-15 | 2018-11-16 | 北京市商汤科技开发有限公司 | The target object segmentation of image and training method and device, equipment, medium, product |
-
2018
- 2018-12-13 CN CN201811523003.8A patent/CN109635813B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8432410B1 (en) * | 2006-07-28 | 2013-04-30 | Nvidia Corporation | 3D graphics API extension for a shared exponent image format |
| CN101778275A (en) * | 2009-01-09 | 2010-07-14 | 深圳市融创天下科技发展有限公司 | Image processing method of self-adaptive time domain and spatial domain resolution ratio frame |
| US20170019684A1 (en) * | 2014-05-20 | 2017-01-19 | Here Global B.V. | Predictive value data set compression |
| CN107977598A (en) * | 2016-10-21 | 2018-05-01 | 三星电子株式会社 | Method and apparatus for identifying facial expression |
| CN107481188A (en) * | 2017-06-23 | 2017-12-15 | 珠海经济特区远宏科技有限公司 | A kind of image super-resolution reconstructing method |
| CN107357899A (en) * | 2017-07-14 | 2017-11-17 | 吉林大学 | Based on the short text sentiment analysis method with product network depth autocoder |
| CN108280452A (en) * | 2018-01-26 | 2018-07-13 | 深圳市唯特视科技有限公司 | A kind of image, semantic label correction method based on parallel network framework |
| CN108510560A (en) * | 2018-04-11 | 2018-09-07 | 腾讯科技（深圳）有限公司 | Image processing method, device, storage medium and computer equipment |
| CN108830221A (en) * | 2018-06-15 | 2018-11-16 | 北京市商汤科技开发有限公司 | The target object segmentation of image and training method and device, equipment, medium, product |
Non-Patent Citations (4)
| Title |
|---|
| AARON VAN DEN OORD 等: "Representation Learning with Contrastive Predictive Coding", 《ARXIV》 * |
| BADRINARAYANAN, VIJAY 等: "SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation", 《IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE》 * |
| ZHENGXUE CHENG 等: "Performance Comparison of Convolutional AutoEncoders, Generative Adversarial Networks and Super-Resolution for Image Compression", 《ARXIV》 * |
| 司宁博: "基于卷积神经网络的图像特征提取算法与图像分类问题研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109635813B (en) | 2020-12-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR101282196B1 (en) | Apparatus and method for separating foreground and background of based codebook In a multi-view image | |
| US9639065B2 (en) | Method and electronic device for displaying operating state information | |
| EP3343376B1 (en) | Disk capacity prediction method, device and apparatus | |
| CN111179187B (en) | Single image rain removing method based on cyclic generation countermeasure network | |
| CN104219532B (en) | The method and apparatus for determining interpolation frame between the method in wisp region, frame of video | |
| CN104881415B (en) | The method and apparatus of tree graph generation | |
| CN105512382A (en) | Floor block conversion and combination method and system based on BIM | |
| CN107180117A (en) | Chart recommends method, device and computer equipment | |
| CN115376211B (en) | Lip driving method, lip driving model training method, device and equipment | |
| CN105718848A (en) | Quality evaluation method and apparatus of fingerprint images | |
| CN103813169A (en) | Extensible object notation method and device for use in video coder/decoder | |
| CN114715145B (en) | Trajectory prediction method, device and equipment and automatic driving vehicle | |
| CN103400393A (en) | Image matching method and system | |
| CN109784207B (en) | Face recognition method, device and medium | |
| CN109635813A (en) | A kind of steel rail area image partition method and device | |
| CN109753157A (en) | The gestural control method and device of display screen | |
| CN113269806B (en) | Method, device and processor for measuring blood flow inside blood vessel | |
| CN114627025A (en) | Image denoising method and device, electronic equipment and storage medium | |
| CN109993338A (en) | A kind of link prediction method and device | |
| CN114757855A (en) | Method, device, equipment and storage medium for correcting action data | |
| CN109993078A (en) | Image-recognizing method, device and the equipment of vehicle environment | |
| CN115661449B (en) | Image segmentation and training method and device for image segmentation model | |
| CN105740319B (en) | A kind of method and device for correcting sampled data | |
| CN108806566A (en) | Move display control method, device and electronic equipment | |
| CN114782595B (en) | Three-dimensional engine-based shortest distance track determination method, device and 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 |