CN103345747A - Optimized picture shape feature extraction and structuring description device and method based on horizontal coordinate - Google Patents
Optimized picture shape feature extraction and structuring description device and method based on horizontal coordinate Download PDFInfo
- Publication number
- CN103345747A CN103345747A CN2013102585227A CN201310258522A CN103345747A CN 103345747 A CN103345747 A CN 103345747A CN 2013102585227 A CN2013102585227 A CN 2013102585227A CN 201310258522 A CN201310258522 A CN 201310258522A CN 103345747 A CN103345747 A CN 103345747A
- Authority
- CN
- China
- Prior art keywords
- effective pixel
- pixel points
- point
- image
- picture
- 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
Images
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses an optimized picture shape feature extraction and structuring description device and method based on a horizontal coordinate. According to the scheme, PAL format analog video signals are encoded through a TVP5150 and then are collected by a DM6446 chip, an algorithm processing thread in a codec engine frame is called for achieving processing of video data and extracting of picture shape features, structuring descriptions are conducted on the features, and finally a processed result is uploaded to the Internet through the Ethernet so that users can achieve retrieval. With regard to extraction of the picture shape features, when the structuring descriptions are conducted on the features, first, graying processing is needed, then binarization processing is carried out, edge processing is conducted, afterwards, all positions with effective pixels appearing are recorded into coordinate data, the data are further divided into two lines, one line is used for browsing the shape of a picture and reappearing the original shape of the picture, and the other line is used for describing the shape in a semantization mode. The optimized picture shape feature extraction and structuring description device and method can lower product cost, reduce network data flow and improve real-time performance.
Description
Technical field
The invention belongs to the data image signal processing technology field, related to a kind of method of intelligent video analysis and the realization technology of semantic video, especially relate to a kind of picture shape feature extraction and structural description method.
Background technology
The high speed development of internet has brought great variety for Human's production and life, and in the face of magnanimity information, various advanced persons' retrieval technique has obtained develop rapidly, provides convenience for obtaining effective information.The birth of Internet of Things concept has been injected new vitality to Internet development, its target is to connect into objects all on the earth a huge network, the the most extensive of promotion information shared, yet, information acquisition for Internet of things system is brought in, at present also there is not general sensor can gather the feature of all objects, various sensors all can only be gathered very a spot of object information, the construction networked system will need huge number of sensors like this, also exist the interface compatibility problem between the various sensors, appearing to a certain extent of machine vision solves these problems, video image can obtain information more widely through the digital processing in later stage, and this information acquisition for Internet of things system provides a kind of good method.
Vision system is because its data volume is huge, be subject to the development of memory technology, seem very necessary so from the huge data volume of vision system, extract typical feature, this process of extracting characteristic from video is called the semantization of video, it is the important component part of intelligent video analysis, video content by analysis will be read by machine with forms such as texts, can realize the retrieval to video content like this.And do not need to transmit huge video stream data, and reduced the requirement to the network bandwidth greatly, can guarantee the timely transmission of important information in the video.
At this patent related picture shape feature extraction and structural description, a variety of methods are also arranged at present, mainly can be divided into two class method for expressing, a class is contour feature, another kind of is provincial characteristics.The contour feature of image is primarily aimed at the outer boundary of object, and the provincial characteristics of image is then paid close attention to whole shape area.Two kinds of methods cut both ways, and wherein, the region representation of object is represented with respect to profile because it has comprised the space distribution information of pixel, have higher anti-interference, but calculated amount is bigger.And existing various algorithm major part is based on PC exploitation, and embedded system has been proposed very high request, makes it be difficult to realize.
Leonardo da Vinci's technology of Texas Instrument (TI) is a DSP solution that customizes at digital video application, it is integrated rich video processing module and interface resource, and provide software development framework, be very easy to the exploitation of embedded intelligence video system, make processing system for video break away from traditional PC, move towards to gather front end.
Summary of the invention
The object of the present invention is to provide a kind of picture shape feature extraction and structural description device and method that is applied to embedded system, be used for solving numerous deficiencies that conventional apparatus adopts the PC scheme to bring owing to algorithm complexity, not only reduce the cost of product, also reduce network traffic data, improved real-time.
In order to achieve the above object, the present invention adopts following technical scheme:
A kind of picture shape feature extraction and structural description device that is applied to embedded system, described device comprises: hardware layer, driving layer, inner nuclear layer, application layer and algorithm function layer, described hardware layer adopts DM6446 as master cpu, encoder chip adopts TVP5150, and joins with main control chip DM6446 data; Described driving layer is mainly finished collection of video signal, and the V4L2 that adopts linux system to carry drives; Described inner nuclear layer operation MontaVista linux system is responsible for coordinating each thread; Described application layer is finished each thread task, and it will call the algorithm file of DSP end by algorithm frame; Described algorithm function layer is used for realizing the picture shape Feature Extraction, and this feature is carried out structural description.
Further, described application layer adopts Codec Engine Frame Design algorithm, and wherein ARM end operation LINUX operating system mainly comprises main thread, collecting thread, processing threads, demonstration thread, control thread; DSP end operation BIOS operating system, the processing threads of ARM end will call the algorithms library file among the DSP, execution algorithm.
As second purpose of the present invention, a kind of picture shape feature extraction and structural description method of optimizing based on planimetric coordinates, described method step following steps:
(1) after the analog video signal of pal mode is encoded through TVP5150, gathered by the DM6446 chip;
(2) finish processing to video data by calling algorithm process thread in the codec engine framework, realize the picture shape Feature Extraction, and this feature is carried out structural description;
(3) result that will handle at last is by uploading to the webserver, and the user can visit this server and obtain data resource.
In the preferred embodiment of this programme, described picture shape Feature Extraction method is analyzed at the coordinate of the effective pixel points of bianry image, distribute the computed image shape facility by coordinate, and the original shape of energy fast restore picture shape, it is mainly with after the gray processing processing of digital picture process and two value transforms that collect, by the processing of edge detection operator, extract the framework of image at last again.
Further, the description of described picture shape feature is by semantic and original shape dual mode, and described semantization description has comprised mainly that image ranks maximum is wide, girth, area, rectangle similarity, circular similarity, and image has been carried out quantitative description; It is the effective pixel points of describing image that described original shape is described, the original shape that these points can reproduced image.
Further again, the semantization of described picture shape feature is described with extracting method as follows:
(A1) effective pixel points of extracting data image original shape after the edge treated, effective pixel points refers to gray-scale value value P[i] and the point of [j]=255 (i, j), remaining pixel P[i] [j]=0;
(A2) framework of computed image profile, the pixel of giving up its profile frame inside, analyze 8 neighborhoods of each effective pixel points in the plane respectively, if the pixel in 8 neighborhoods of certain effective pixel points is effective pixel points then this point belongs to the pixel of profile frame inside; If in its 8 neighborhood be not effective pixel points then it is the framework pixel of image profile entirely, and this point coordinate be recorded in two-dimensional array F[F
Num] in [2]; If have only an effective pixel points in 8 neighborhoods of some effective pixel points, this effective pixel points is breakpoint, then enlarge search neighborhood another effective pixel points nearest with it and carry out interpolation arithmetic, fill an effective pixel points in the neighborhood of breakpoint, the framework that makes profile is an enclosed areas;
(A3) the planar coordinate F[F of image profile frame will be described
Num] [2] point set carries out vectorized process, and the planar coordinate points is transformed into the linking point that has direction, and guarantee 2 effective pixel points can only occur in 8 fields of each pixel, if be less than 2, then in step (A2), filled one by method of interpolation; If more than 2 effective pixel points, then before the next linking point of search, analyze in this neighborhood the weight of other effective pixel points except linking point, namely analyze the number of the effective pixel points except close linking point in these pixel neighborhoods of a point, naming a person for a particular job that effective pixel points is many in the neighborhood is used as next linking point;
(A4) for the ranks breadth extreme of shape: at two-dimensional array F[F
Num] [2] middle search (i, j) the minimum i in of having a few
MinWith maximum i
Max, and minimum j
MinWith maximum j
Max, its row maximum line length is j
Max-j
MinBeing listed as maximum col width is i
Max-i
Min
(A5) for the girth of shape: from image profile frame vector point set F ' [F
Num] first point (F ' [1] [0], F ' [1] [1]) beginning of [2], analyze it and whether have at least one to be identical with next row or row coordinate of putting, if do not have, illustrate that then the distance between these two points is
Individual pixel, is analyzed in the order that chains one by one according to vector point, counts distance between two points and is
The number Q of individual pixel
Num, can calculate girth;
(A6) for the area of shape: with F ' [F
Num] the row coordinate of having a few on certain row in [2] sorts, calculate then from the row coordinate of minimum and begin pixel number to next row coordinate, the point that also has other on this row then calculates from the row coordinate of next one point again and begins to the pixel number of next but one point range coordinate.According to this process with F ' [F
Num] all line numbers have all been added up in [2], calculate total pixel number S
Num, can calculate area;
(A7) for rectangle similarity σ: the rectangle similarity is used for weighing the shape and the similarity degree that comprises the minimum rectangle of its all pixels, computing formula of image
(A8) for circular similarity δ: adopt density to measure the circularity of picture shape, namely girth (L) square with the ratio of area (S):
Further again, the original shape of described picture shape feature is described with the original shape extracting method according to the two-dimensional array F[Fnum that comprises the effective pixel points coordinate] [2] go back the original image original shape.
Scheme provided by the invention has realized carrying out picture shape feature extraction and structural description at the DM6446 platform, has comparatively wide applications.
Simultaneously, scheme provided by the invention has adopted distinct methods at the different characteristic of picture shape, reach optimal effectiveness, the algorithm that adopts simultaneously is based on planimetric coordinates, the core of its algorithm is that logic is judged, need not to carry out too many multiplication and division computing, this will greatly improve the efficient of system for the fixed point embedded system, reduce the expense of CPU.
Description of drawings
Further specify the present invention below in conjunction with the drawings and specific embodiments.
Fig. 1 is that the system of this device constitutes block diagram;
Fig. 2 is the operation logic figure of this device;
Fig. 3 is the development process figure of this device;
Fig. 4 optimizes the algorithm flow block diagram based on planimetric coordinates;
Fig. 5 is coordinate points set vector synoptic diagram.
Embodiment
For technological means, creation characteristic that the present invention is realized, reach purpose and effect is easy to understand, below in conjunction with concrete diagram, further set forth the present invention.
Picture shape feature extraction and the structural description device of optimizing based on planimetric coordinates provided by the invention, it can replace currently marketed PC product, when reducing volume and cost, reduce the network traffics of system, can be applied to the video wireless acquisition terminal, improve the intelligent level of gathering front end.
Referring to Fig. 1, it is depicted as the system chart of this device.Hence one can see that, and whole device mainly is made of hardware layer, driving layer, inner nuclear layer, application layer and algorithm function layer.
Adopt DM6446 as master cpu in the hardware layer, adopt TVP5150 as coding chip;
Drive layer and mainly finish collection of video signal, the V4L2 that adopts linux system to carry drives;
Inner nuclear layer operation MontaVista linux system is responsible for coordinating each thread;
Application layer is finished each thread task, and it will call the algorithm file of DSP end by algorithm frame; This application layer adopts Codec Engine Frame Design algorithm, and wherein ARM end operation LINUX operating system mainly comprises main thread, collecting thread, processing threads, demonstration thread, control thread; DSP end operation BIOS operating system, the processing threads of ARM end will call the algorithms library file among the DSP, execution algorithm;
The algorithm function layer is used for realizing the picture shape Feature Extraction, and this feature is carried out structural description.In the present invention, the description of picture shape feature is described by semantic and original shape dual mode: the semantization description has comprised mainly that image ranks maximum is wide, girth, area, rectangle similarity, circular similarity, and image has been carried out quantitative description; It mainly is the effective pixel points of describing image that original shape is described, the original shape that these points can reproduced image.
Referring to Fig. 2, picture shape feature extraction and the structural description device based on the planimetric coordinates optimization that form thus, it is when operation, after the analog video signal of pal mode is encoded through TVP5150, gathered by the DM6446 chip, finish processing to video data by calling algorithm process thread in the codec engine framework, upload to the webserver by cable network or wireless transmission method at last, the user can visit this server and obtain data resource.The algorithm function layer is realized the picture shape Feature Extraction by various algoritic modules, and this feature is carried out structural description.The picture shape feature of this unit describe mainly comprises: the ranks maximum of image is wide, girth, area, rectangle similarity, circular similarity and shape are browsed.Obtain the image of object by the candid photograph program after, at first its gray processing is handled, carry out binary conversion treatment again, carry out edge treated then, then all positions that effective pixel points occurs are recorded as coordinate data, again these data are divided into two-way: the one tunnel is used for picture shape browses, and is used for reproduced image shape original shape, and one the tunnel is used for semantization describes shape.
For above-mentioned technical scheme, the present invention illustrates the implementation process that it is whole by a concrete example.
Referring to Fig. 3, it is depicted as based on the picture shape feature extraction of planimetric coordinates optimization and the development process figure of structural description device.
A1, design hardware circuit principle figure, main control chip adopts DM6446, and encoder chip adopts TVP5150, and ethernet control chip is selected LXT971ALC for use, adopts two MT47H64M16BT to constitute the DDR of 128M.Making sheet post debugging monoblock circuit board is primarily aimed at power-supply system, video acquisition function, storage system, network function etc.
A2, development board level support package comprise relevant driving file, Uboot file and UBL file etc.
A3, design function algorithm as shown in Figure 4, mainly may further comprise the steps:
A3.1, utilize the candid photograph program from the Buffer that gathers, to obtain frame data, also a width of cloth is contained the view data of target, and these data are carried out gray scale is handled, two-value is handled and edge treated, can directly call the IMGLIB built-in function of TI.
The framework of A3.2, extraction image profile.The picture shape Feature Extraction mainly with after the gray processing processing of digital picture process and two value transforms that collect, again by the processing of edge detection operator, is extracted the framework of image at last.Picture shape Feature Extraction algorithm is mainly paid close attention to image distribution in the plane, so the designed algorithm of this device is a kind of algorithm based on planimetric coordinates, core concept is to analyze at the coordinate of the effective pixel points of bianry image, distribute the computed image shape facility by coordinate, and the original shape of energy fast restore picture shape.
For this reason, the effective pixel points of this method extracting data image original shape after the edge treated, effective pixel points refers to gray-scale value value P[i] and the point of [j]=255 (i, j), remaining pixel P[i] [j]=0.The framework of computed image profile is given up the pixel of its profile frame inside.Analyze 8 neighborhoods of each effective pixel points in the plane respectively, if the pixel in 8 neighborhoods of certain effective pixel points is entirely for effective pixel points then this point belongs to the pixel of profile frame inside, delete this point, otherwise it is the framework point of image profile, and with its coordinate record at two-dimensional array F[F
Num] in [2], the main contents that this array will be described as the picture shape original shape.If have only an effective pixel points in 8 neighborhoods of certain point, illustrate that this point is breakpoint, need expansion search neighborhood and another effective pixel points nearest with it to carry out interpolation arithmetic, fill the effective pixel points of interpolation gained in the neighborhood of breakpoint, the framework that makes profile is an enclosed areas.
A3.3, with coordinate point set F[F
Num] [2] carry out vectorized process, selecting a point is starting point, search for nearest in its 8 a neighborhood point as next point, this names a person for a particular job and done by note (F ' [1] [0], F ' [1] [1]), its starting point is made vector point (F ' [0] [0], F ' [0] [1]) by note, by that analogy the planar coordinate points is transformed into the linking point that has direction, the order of linking point is by vector point set F ' [F
Num] F in [2]
NumIdentify.But need in 8 fields of each pixel of assurance 2 effective pixel points can only appear, if more than 2 effective pixel points, then need be before the next linking point of search, analyze in this neighborhood the weight of other effective pixel points except linking point, namely analyze total number of the effective pixel points except close linking point in these pixel neighborhoods of a point, naming a person for a particular job that effective pixel points is many in the neighborhood is used as next linking point, and remaining point is rejected.As shown in Figure 5, after the vector chain pointed to some B by some A, the point that some B can point to its 8 field had H, E, C, D, and wherein the neighborhood of E has H, C, F, I, G totally 5 effective pixel points, and many than E, C, D will be so the vector chain will point to a some E by a B, by that analogy.
The ranks breadth extreme of A3.4, calculating figure.At two-dimensional array F[F
Num] [2] middle search (i, j) the minimum i in of having a few
MinWith maximum i
Max, and minimum j
MinWith maximum j
Max, its row breadth extreme is j
Max-j
MinThe row breadth extreme is i
Max-i
Min
A3.5, calculating girth.From image profile frame vector point set F ' [F
Num] first point (F ' [1] [0], F ' [1] [1]) beginning of [2], analyze it and whether have at least one to be identical with next row or row coordinate of putting, if do not have, illustrate that then the distance between these two points is
Individual pixel, is analyzed in the order that chains one by one according to vector point, counts distance between two points and is
The number Q of individual pixel
Num, then its girth is:
Wherein: F
dBe the corresponding physical length of unit picture element point.
A3.6, reference area.Process is as follows: with F ' [F
Num] the row coordinate of having a few on certain row in [2] sorts, calculate then from the row coordinate of minimum and begin pixel number to next row coordinate, the point that also has other on this row then calculates from the row coordinate of next one point again and begins to the pixel number of next but one point range coordinate.According to this process with F ' [F
Num] all line numbers have all been added up in [2], calculate total pixel number S
Num, area then
A3.7, calculating rectangle similarity σ.The rectangle similarity is used for weighing the shape and the similarity degree that comprises the minimum rectangle of its all pixels, computing formula of image
A3.8, circular similarity δ.Adopt density to measure the circularity of picture shape, namely girth (L) square with the ratio of area (S):
A4, above-mentioned algorithm is transplanted on the CCS3.3 development environment, a newly-built library file engineering compiles this algorithm, generates the .lib file.
A5, enter the Linux development environment, the required DVSDK support package of exploitation DM6446 is installed, the line correlation of going forward side by side configuration is as path setting, system parameter setting etc.
A6, in Codec Engine framework, add the support to this algorithms library file, in the cmd file, add above-mentioned .lib algorithm file.And in algorithm frame, call the packaged algorithmic function of this algorithms library file.This function is input as the Buffer pointer of video acquisition, is output as the structure pointer of describing the image aspects feature, and its inside has comprised the result that semantization is described and original shape is described.
By above-mentioned example as can be known, the scheme of picture shape feature extraction provided by the invention and structural description, the method that this programme is optimized based on planimetric coordinates, this method judges to replace mathematical operation by adopting logic, thereby improved efficient greatly, so the disclosed method of this patent is applicable to multiple embedded platform, also having above-mentioned embodiment only is a kind of implementation, but is not limited to this.
More than show and described ultimate principle of the present invention, principal character and advantage of the present invention.The technician of the industry should understand; the present invention is not restricted to the described embodiments; that describes in above-described embodiment and the instructions just illustrates principle of the present invention; without departing from the spirit and scope of the present invention; the present invention also has various changes and modifications, and these changes and improvements all fall in the claimed scope of the invention.The claimed scope of the present invention is defined by appending claims and equivalent thereof.
Claims (7)
1. picture shape feature extraction and structural description device that is applied to embedded system, it is characterized in that, described device comprises: hardware layer, driving layer, inner nuclear layer, application layer and algorithm function layer, described hardware layer adopts DM6446 as master cpu, encoder chip adopts TVP5150, and joins with main control chip DM6446 data; Described driving layer is mainly finished collection of video signal, and the V4L2 that adopts linux system to carry drives; Described inner nuclear layer operation MontaVista linux system is responsible for coordinating each thread; Described application layer is finished each thread task, and it will call the algorithm file of DSP end by algorithm frame; Described algorithm function layer is used for realizing the picture shape Feature Extraction, and this feature is carried out structural description.
2. a kind of picture shape feature extraction and structural description device that is applied to embedded system according to claim 1, it is characterized in that, described application layer adopts Codec Engine Frame Design algorithm, wherein ARM end operation LINUX operating system mainly comprises main thread, collecting thread, processing threads, demonstration thread, control thread; DSP end operation BIOS operating system, the processing threads of ARM end will call the algorithms library file among the DSP, execution algorithm.
3. picture shape feature extraction and structural description method of optimizing based on planimetric coordinates is characterized in that described method step following steps:
(1) after the analog video signal of pal mode is encoded through TVP5150, gathered by the DM6446 chip;
(2) finish processing to video data by calling algorithm process thread in the codec engine framework, realize the picture shape Feature Extraction, and this feature is carried out structural description;
(3) result that will handle at last is by uploading to the webserver, and the user can visit this server and obtain data resource.
4. a kind of picture shape feature extraction and structural description method of optimizing based on planimetric coordinates according to claim 3, it is characterized in that, described picture shape Feature Extraction method is analyzed at the coordinate of the effective pixel points of bianry image, distribute the computed image shape facility by coordinate, and the original shape of energy fast restore picture shape, it is mainly with after the gray processing processing of digital picture process and two value transforms that collect, by the processing of edge detection operator, extract the framework of image at last again.
5. according to claim 3 or 4 described a kind of picture shape feature extraction and structural description methods of optimizing based on planimetric coordinates, it is characterized in that, the description of described picture shape feature is by semantic and original shape dual mode, described semantization describe comprised mainly that image ranks maximum is wide, girth, area, rectangle similarity, circular similarity, image has been carried out quantitative description; It is the effective pixel points of describing image that described original shape is described, the original shape that these points can reproduced image.
6. a kind of picture shape feature extraction and structural description method of optimizing based on planimetric coordinates according to claim 5 is characterized in that, the semantization description of described picture shape feature is as follows with extracting method:
(A1) effective pixel points of extracting data image original shape after the edge treated, effective pixel points refers to gray-scale value value P[i] and the point of [j]=255 (i, j), remaining pixel P[i] [j]=0;
(A2) framework of computed image profile, the pixel of giving up its profile frame inside, analyze 8 neighborhoods of each effective pixel points in the plane respectively, if the pixel in 8 neighborhoods of certain effective pixel points is effective pixel points then this point belongs to the pixel of profile frame inside; If in its 8 neighborhood be not effective pixel points then it is the framework pixel of image profile entirely, and this point coordinate be recorded in two-dimensional array F[F
Num] in [2]; If have only an effective pixel points in 8 neighborhoods of some effective pixel points, this effective pixel points is breakpoint, then enlarge search neighborhood another effective pixel points nearest with it and carry out interpolation arithmetic, fill an effective pixel points in the neighborhood of breakpoint, the framework that makes profile is an enclosed areas;
(A3) the planar coordinate F[F of image profile frame will be described
Num] [2] point set carries out vectorized process, and the planar coordinate points is transformed into the linking point that has direction, and guarantee 2 effective pixel points can only occur in 8 fields of each pixel, if be less than 2, then in step (A2), filled one by method of interpolation; If more than 2 effective pixel points, then before the next linking point of search, analyze in this neighborhood the weight of other effective pixel points except linking point, namely analyze the number of the effective pixel points except close linking point in these pixel neighborhoods of a point, naming a person for a particular job that effective pixel points is many in the neighborhood is used as next linking point;
(A4) for the ranks breadth extreme of shape: at two-dimensional array F[F
Num] [2] middle search (i, j) the minimum i in of having a few
MinWith maximum i
Max, and minimum j
MinWith maximum j
Max, its row maximum line length is j
Max-j
MinBeing listed as maximum col width is i
Max-i
Min
(A5) for the girth of shape: from image profile frame vector point set F ' [F
Num] first point (F ' [1] [0], F ' [1] [1]) beginning of [2], analyze it and whether have at least one to be identical with next row or row coordinate of putting, if do not have, illustrate that then the distance between these two points is
Individual pixel, is analyzed in the order that chains one by one according to vector point, counts distance between two points and is
The number Q of individual pixel
Num, can calculate girth;
(A6) for the area of shape: with F ' [F
Num] the row coordinate of having a few on certain row in [2] sorts, calculate then from the row coordinate of minimum and begin pixel number to next row coordinate, the point that also has other on this row then calculates from the row coordinate of next one point again and begins to the pixel number of next but one point range coordinate.According to this process with F ' [F
Num] all line numbers have all been added up in [2], calculate total pixel number S
Num, can calculate area;
(A7) for rectangle similarity σ: the rectangle similarity is used for weighing the shape and the similarity degree that comprises the minimum rectangle of its all pixels, computing formula of image
(A8) for circular similarity δ: adopt density to measure the circularity of picture shape, namely girth (L) square with the ratio of area (S):
7. a kind of picture shape feature extraction and structural description method of optimizing based on planimetric coordinates according to claim 5, it is characterized in that the original shape of described picture shape feature is described with the original shape extracting method according to the two-dimensional array F[Fnum that comprises the effective pixel points coordinate] [2] go back the original image original shape.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2013102585227A CN103345747A (en) | 2013-06-26 | 2013-06-26 | Optimized picture shape feature extraction and structuring description device and method based on horizontal coordinate |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2013102585227A CN103345747A (en) | 2013-06-26 | 2013-06-26 | Optimized picture shape feature extraction and structuring description device and method based on horizontal coordinate |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103345747A true CN103345747A (en) | 2013-10-09 |
Family
ID=49280540
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2013102585227A Pending CN103345747A (en) | 2013-06-26 | 2013-06-26 | Optimized picture shape feature extraction and structuring description device and method based on horizontal coordinate |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103345747A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103729181A (en) * | 2013-12-27 | 2014-04-16 | 国电南京自动化股份有限公司 | Software modular method based on CSS platform |
| CN104951785A (en) * | 2015-06-05 | 2015-09-30 | 中国民用航空总局第二研究所 | Method for evaluating feature description processes |
| CN105677107A (en) * | 2016-01-11 | 2016-06-15 | 福州瑞芯微电子股份有限公司 | Method and system for optimizing screen handwriting following performance |
| CN110390568A (en) * | 2019-07-03 | 2019-10-29 | 北京旋极百旺科技有限公司 | A kind of billing system and its loading method based on layer-stepping principle |
| CN112270370A (en) * | 2020-11-06 | 2021-01-26 | 北京环境特性研究所 | Vehicle apparent damage assessment method |
| CN112750143A (en) * | 2020-12-02 | 2021-05-04 | 上海海洋大学 | Extreme method-based method for extracting morphological characteristics of tender soft fish |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6393150B1 (en) * | 1998-12-04 | 2002-05-21 | Eastman Kodak Company | Region-based image binarization system |
| CN101447017A (en) * | 2008-11-27 | 2009-06-03 | 浙江工业大学 | Method and system for quickly identifying and counting votes on the basis of layout analysis |
| CN101827266A (en) * | 2010-04-01 | 2010-09-08 | 公安部第三研究所 | Network video server with video structural description function and method for implementing video analysis description by using same |
| CN102724485A (en) * | 2012-06-26 | 2012-10-10 | 公安部第三研究所 | Device and method for performing structuralized description for input audios by aid of dual-core processor |
-
2013
- 2013-06-26 CN CN2013102585227A patent/CN103345747A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6393150B1 (en) * | 1998-12-04 | 2002-05-21 | Eastman Kodak Company | Region-based image binarization system |
| CN101447017A (en) * | 2008-11-27 | 2009-06-03 | 浙江工业大学 | Method and system for quickly identifying and counting votes on the basis of layout analysis |
| CN101827266A (en) * | 2010-04-01 | 2010-09-08 | 公安部第三研究所 | Network video server with video structural description function and method for implementing video analysis description by using same |
| CN102724485A (en) * | 2012-06-26 | 2012-10-10 | 公安部第三研究所 | Device and method for performing structuralized description for input audios by aid of dual-core processor |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103729181A (en) * | 2013-12-27 | 2014-04-16 | 国电南京自动化股份有限公司 | Software modular method based on CSS platform |
| CN104951785A (en) * | 2015-06-05 | 2015-09-30 | 中国民用航空总局第二研究所 | Method for evaluating feature description processes |
| CN104951785B (en) * | 2015-06-05 | 2017-12-08 | 中国民用航空总局第二研究所 | A kind of method weighed to character description method |
| CN105677107A (en) * | 2016-01-11 | 2016-06-15 | 福州瑞芯微电子股份有限公司 | Method and system for optimizing screen handwriting following performance |
| CN105677107B (en) * | 2016-01-11 | 2018-07-17 | 福州瑞芯微电子股份有限公司 | The optimization method and system of the hand-written followability of screen |
| CN110390568A (en) * | 2019-07-03 | 2019-10-29 | 北京旋极百旺科技有限公司 | A kind of billing system and its loading method based on layer-stepping principle |
| CN112270370A (en) * | 2020-11-06 | 2021-01-26 | 北京环境特性研究所 | Vehicle apparent damage assessment method |
| CN112270370B (en) * | 2020-11-06 | 2023-06-02 | 北京环境特性研究所 | Vehicle apparent damage assessment method |
| CN112750143A (en) * | 2020-12-02 | 2021-05-04 | 上海海洋大学 | Extreme method-based method for extracting morphological characteristics of tender soft fish |
| CN112750143B (en) * | 2020-12-02 | 2024-04-26 | 上海海洋大学 | Method for extracting morphological characteristics of stem flexible fish based on extremum method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103345747A (en) | Optimized picture shape feature extraction and structuring description device and method based on horizontal coordinate | |
| CN110378222A (en) | A kind of vibration damper on power transmission line target detection and defect identification method and device | |
| KR20190052893A (en) | Method and apparatus for preprocessing an operation of neural network | |
| CN111126287B (en) | Remote sensing image dense target deep learning detection method | |
| CN114943876A (en) | Cloud and cloud shadow detection method and device for multi-level semantic fusion and storage medium | |
| CN104217417B (en) | A kind of method and device of video multi-target tracking | |
| CN111881238A (en) | Lightweight three-dimensional data construction method and medium suitable for Web end and electronic device | |
| CN108764247A (en) | Deep learning object detecting method and device based on dense connection | |
| CN114049545B (en) | Typhoon intensity determining method, system, equipment and medium based on point cloud voxels | |
| CN112541459A (en) | Crowd counting method and system based on multi-scale perception attention network | |
| CN112819066A (en) | Res-UNet single tree species classification technology | |
| CN105931256A (en) | CUDA (compute unified device architecture)-based large-format remote sensing image fast segmentation method | |
| Zhou et al. | A streaming framework for seamless building reconstruction from large-scale aerial lidar data | |
| Ding et al. | Deep Learning Approach for Estimating Workability of Self‐Compacting Concrete from Mixing Image Sequences | |
| CN113205018A (en) | High-resolution image building extraction method based on multi-scale residual error network model | |
| CN112819796A (en) | Tobacco shred foreign matter identification method and equipment | |
| CN107888970A (en) | Method for processing video frequency, device, embedded device and storage medium | |
| CN116434012A (en) | Lightweight cotton boll detection method and system based on edge perception | |
| CN115719445A (en) | Seafood identification method based on deep learning and raspberry type 4B module | |
| CN107767486A (en) | A kind of storage method and device of the end-to-end automatic Pilot data based on video format | |
| CN113888505B (en) | Natural scene text detection method based on semantic segmentation | |
| Lin et al. | Integrated circuit board object detection and image augmentation fusion model based on YOLO | |
| CN112668416A (en) | Real-time analysis and identification method for distribution network inspection video of unmanned aerial vehicle and embedded inspection device | |
| CN105093283A (en) | Three-dimensional observation system surface element attribute multi-thread rapid display method | |
| CN115063478B (en) | Fruit positioning method, system, equipment and medium based on RGB-D camera and visual positioning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20131009 |