CN109034173A - Target object choosing method and device - Google Patents
Target object choosing method and device Download PDFInfo
- Publication number
- CN109034173A CN109034173A CN201710427979.4A CN201710427979A CN109034173A CN 109034173 A CN109034173 A CN 109034173A CN 201710427979 A CN201710427979 A CN 201710427979A CN 109034173 A CN109034173 A CN 109034173A
- Authority
- CN
- China
- Prior art keywords
- image
- target object
- target
- target area
- load
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/40—Software arrangements specially adapted for pattern recognition, e.g. user interfaces or toolboxes therefor
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
- G06F3/048—Interaction techniques based on graphical user interfaces [GUI]
- G06F3/0484—Interaction techniques based on graphical user interfaces [GUI] for the control of specific functions or operations, e.g. selecting or manipulating an object, an image or a displayed text element, setting a parameter value or selecting a range
- G06F3/04842—Selection of displayed objects or displayed text elements
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
- G06F3/048—Interaction techniques based on graphical user interfaces [GUI]
- G06F3/0484—Interaction techniques based on graphical user interfaces [GUI] for the control of specific functions or operations, e.g. selecting or manipulating an object, an image or a displayed text element, setting a parameter value or selecting a range
- G06F3/04845—Interaction techniques based on graphical user interfaces [GUI] for the control of specific functions or operations, e.g. selecting or manipulating an object, an image or a displayed text element, setting a parameter value or selecting a range for image manipulation, e.g. dragging, rotation, expansion or change of colour
-
- 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
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2203/00—Indexing scheme relating to G06F3/00 - G06F3/048
- G06F2203/048—Indexing scheme relating to G06F3/048
- G06F2203/04806—Zoom, i.e. interaction techniques or interactors for controlling the zooming operation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Human Computer Interaction (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- User Interface Of Digital Computer (AREA)
Abstract
The present invention provides a kind of target object choosing method and devices, wherein this method comprises: Auto load Images;By template trained in advance, target detection is carried out to the image of load;In the case where detecting that mouse is moved to the target area detected, the automatic amplification of hypergeometric example mapping is carried out to the target area detected in image, in the case where there is the target area being not detected in the picture, in response to the clicking operation to the target area, the automatic amplification of hypergeometric example mapping is carried out to the target area being not detected;Automatic amplified region is mapped in hypergeometric example and shows the key point in the target area, and seeks the boundary rectangle of each key point, using boundary rectangle as target object;The target object is renamed and recorded, and records coordinate of the target object in load image.The embodiment of the present invention solves existing target object and chooses the excessively cumbersome technical problem of process, has reached the technical effect for being simple and efficient and realizing that target object is chosen.
Description
Technical field
The present invention relates to machine recognition technical field, in particular to a kind of target object choosing method and device.
Background technique
Currently, typically specially looking for some to utilize the generation of great amount of images sample database in machine learning
Photoshop or other openings of screenshot tool one by one, then, by target ROI (Region Of Interest, sense
Interest region) manually from every image cut come out and manually save, renaming.
However, also need to handle section target image come out in many cases, such as: it is carried out in original image
Ratio expansion etc., at this moment also needs to record coordinate of the target ROI in original image.Generally, it while artificial screenshot, utilizes
Photoshop get the coordinate of target ROI and it is manually recorded get off, process is relatively complicated.
But this mode, in the case where there is image pattern demands tens of thousands of, that hundreds of thousands is even more, it will expend big
Manpower, time and the financial resources of amount.
For this problem, effective solution mode is not yet proposed at present.
Summary of the invention
The embodiment of the invention provides a kind of target object choosing methods, realize that target object is chosen to reach to be simple and efficient
Purpose, this method comprises:
Auto load Images;
By template trained in advance, target detection is carried out to the image of load;
In the case where detecting that mouse is moved to the target area detected, to the target area detected in image into
The automatic amplification of row hypergeometric example mapping, in the case where there is the target area being not detected in the picture, in response to the target
The clicking operation in region carries out the automatic amplification of hypergeometric example mapping to the target area being not detected;
Automatic amplified region is mapped in hypergeometric example and shows the key point in the target area, and seeks each key point
Boundary rectangle, using boundary rectangle as target object;
The target object is renamed and recorded, and records coordinate of the target object in load image.
In one embodiment, the target object is being renamed and is being recorded, and recording the target object
After coordinate in load image, the method also includes:
Perspective transform and the fine tuning of low-angle random direction are carried out to the target object, obtain converting target pair adjusted
The image of elephant;
The ratio of width to height expansion is carried out to the image for converting target object adjusted.
In one embodiment, after Auto load Images, the method also includes:
Determine whether the image of load needs to rotate;
In the case where determining that needs rotate, successively the image clockwise of load is rotated by 90 °, until by image rotation
Just;
The original image for rotating positive image and replacing load is stored.
In one embodiment, in response to carrying out clicking operation to the target area, to the target being not detected
Region carries out the automatic amplification of hypergeometric example mapping, comprising:
Determine clicking operation corresponding click location in the picture;
According to click location, magnification region is determined;
The position that magnification region is replaced to image, is shown in control, wherein the size of control is fixed.
In one embodiment, according to click location, magnification region is determined, comprising:
Divide an image into nine regions, wherein nine regions include: the upper left corner 1/4 of present image, it is current
The upper right corner 1/4 of image, the lower right corner 1/4 of present image, the lower left corner 1/4 of present image, upper 1/4 in present image, when
Lower 1/4 in preceding image, left the 1/4 of present image, in the right side of present image 1/4, present image centre 1/4;
Select a region as region to be amplified from nine regions.
In one embodiment, Auto load Images, comprising:
Load the catalogue of the video frame or image data that capture by camera;
Automatically it detects frame under the catalogue or whether image is processed;
If untreated, successively untreated frame or image are loaded onto Image control.
The embodiment of the invention also provides a kind of target object selecting devices, realize target object choosing to reach to be simple and efficient
The purpose taken, the device include:
Loading module is used for Auto load Images;
Detection module, for carrying out target detection to the image of load by template trained in advance;
Amplification module, in the case where detecting that mouse is moved to the target area detected, to being detected in image
The target area arrived carries out the automatic amplification of hypergeometric example mapping, in the case where there is the target area being not detected in the picture,
In response to the clicking operation to the target area, the automatic amplification of hypergeometric example mapping is carried out to the target area being not detected;
Display module shows the key point in the target area for mapping automatic amplified region in hypergeometric example, and
The boundary rectangle for seeking each key point, using boundary rectangle as target object;
Logging modle for the target object to be renamed and recorded, and records the target object and is loading
Coordinate in image.
In one embodiment, above-mentioned apparatus further include: conversion module, for being ordered to the target object again
Name simultaneously records, and after recording coordinate of the target object in load image, carries out perspective transform to the target object
It is finely tuned with low-angle random direction, obtains the image for converting target object adjusted;To transformation target object adjusted
Image carries out the ratio of width to height expansion.
In one embodiment, the amplification module includes:
First determination unit, for determining clicking operation corresponding click location in the picture;
Second determination unit, for determining magnification region according to click location;
Display unit is shown in control for magnification region to be replaced to the position of image, wherein control it is big
Small is fixed.
In one embodiment, second determination unit includes:
Subelement is divided, for dividing an image into nine regions, wherein nine regions include: present image
The upper left corner 1/4, the upper right corner 1/4 of present image, the lower right corner 1/4 of present image, the lower left corner 1/4 of present image, current figure
In picture lower 1/4 in upper 1/4, present image, left the 1/4 of present image, 1/4, present image in the right side of present image
Centre 1/4;
Subelement is selected, for selecting a region as region to be amplified from nine regions.
In embodiments of the present invention, a kind of target object choosing method and device are provided, can with Auto load Images, from
Dynamic detected target object, automatic amplification target object, and automatically record the position of key point, to target object carry out renaming and
It is automatically stored, so that solving existing target object chooses the excessively cumbersome technical problem of process, has reached and be simple and efficient realization
The technical effect that target object is chosen.
Detailed description of the invention
The drawings described herein are used to provide a further understanding of the present invention, constitutes part of this application, not
Constitute limitation of the invention.In the accompanying drawings:
Fig. 1 is the method flow diagram of target object choosing method according to an embodiment of the present invention;
Fig. 2 is the method flow diagram of the specific embodiment of target object choosing method according to an embodiment of the present invention;
Fig. 3 is the structural block diagram of target object selecting device according to an embodiment of the present invention.
Specific embodiment
To make the objectives, technical solutions, and advantages of the present invention clearer, right below with reference to embodiment and attached drawing
The present invention is described in further details.Here, exemplary embodiment and its explanation of the invention is used to explain the present invention, but simultaneously
It is not as a limitation of the invention.
In order to which the application is better described, below to this application involves some explanations of nouns it is as follows:
1) target ROI (Region Of Interest, area-of-interest), refer in piece image target (such as: vehicle
Board, face etc.) key point seek a boundary rectangle, then, according to demand expand boundary rectangle the ratio of width to height, finally obtain
The rectangular area of proper ratio be just called target ROI;
2) ClickTools, tool title;
3) hypergeometric example maps, and carries out target ROI detection to the image currently loaded using existing trained template, detects
Object can amplify automatically when mouse is put up, and with target ROI original wide high proportion amplification display, this process
Just it is called the mapping of hypergeometric example;
4) perspective transform has some samples (such as: license plate sample) due to artificial shooting angle problem, so that be truncated to
License plate is very oblique, and a license plate ratio is about 3.14:1, and the boundary rectangle ratio of some sample license plates is less than 2:1, these samples
It loses very unfortunately, it is first just subjected to perspective transform, its drawing calculation is allowed to become the rectangle of regular ratio.Target in image
Four key points composition of (such as: license plate) is an irregular quadrangle, this irregular quadrangle is passed through four
The Mapping and Converting of a coordinate points calculates, and projection mapping is called quadrangle perspective transform at a regular rectangular shape, this transformation;
5) low-angle random direction is finely tuned, the rectangle of perspective transform result rule, but training sample (such as: license plate) it needs
There is the inclined polymorphic type sample of all directions all angles, just the higher template of verification and measurement ratio can be trained, in order to rationally utilize
The license plate sample of this fragmentary perspective transformation, generates -15~15 degree of angles at random, is carried out between -15~15 at any angle
This process is known as the fine tuning of low-angle random direction by rotary fine adjustment, available good sample.
When for existing acquisition target ROI as training sample, need to expend a large amount of manpowers, time, material resources
Problem, inventor in view of can provide it is a kind of automatic carry out picture load, carry out recongnition of objects automatically, and automatically into
The mode of row amplification and key point identification and name.As shown in Figure 1, the target object choosing method may include:
Step 101: Auto load Images;
Step 102: by template trained in advance, target detection being carried out to the image of load;
Step 103: in the case where detecting that mouse is moved to the target area detected, to the mesh detected in image
It marks region and carries out the automatic amplification of hypergeometric example mapping, in the case where there is the target area being not detected in the picture, in response to
To the clicking operation of the target area, the automatic amplification of hypergeometric example mapping is carried out to the target area being not detected;
Step 104: mapping automatic amplified region in hypergeometric example and show the key point in the target area, and seek each
The boundary rectangle of key point, using boundary rectangle as target object;
Step 105: the target object being renamed and recorded, and records the target object in load image
Coordinate.
In one embodiment, after above-mentioned steps 105, can also to the target object carry out perspective transform and
The fine tuning of low-angle random direction, obtains the image for converting target object adjusted;To the figure for converting target object adjusted
As carrying out the ratio of width to height expansion, to obtain available and diversified sample.
In view of existing acquisition image sometimes angle be it is of problems, therefore, after Auto load Images,
It can determine whether the image of load needs to rotate;In the case where determining that needs rotate, successively to the image clockwise of load
It is rotated by 90 °, until just by image rotation;The original image for rotating positive image and replacing load is stored.So as to
Positive image after being corrected, the original image as subsequent processing.
In order to enable operator can see the specific appearance and key point therein of target ROI, figure can control
As being amplified automatically, when being amplified automatically, clicking operation corresponding click location in the picture can be first determined;
Then, it according to click location, determines magnification region, and magnification region is replaced to the position of image, shown in control,
In, the size of control is fixed.Wherein, the size of Image control is fixed, for example, being exactly when extracting image
Show that whole image exactly shows amplification when carrying out image amplification in the Image control by the Image control
Region, that is, amplification scale it is bigger, the region of the picture shown in Image control is smaller, but details show it is bigger.
In one embodiment, it according to click location, determines magnification region, may include: to divide an image into nine
Region, wherein nine regions include: the upper left corner 1/4 of present image, the upper right corner 1/4 of present image, present image
The lower right corner 1/4, the lower left corner 1/4 of present image, lower 1/4, present image in upper 1/4, present image in present image
Left 1/4, in the right side of present image 1/4, present image centre 1/4;Selected from nine regions a region as to
Magnification region.
In above-mentioned steps 101, the catalogue of the video frame or image data that capture by camera can be loaded, it is automatic to examine
It surveys frame under the catalogue or whether image is processed, if untreated, image successively is loaded onto untreated frame or image
In control, to realize the identifying processing to every image in catalogue.
The above method is illustrated below with reference to a specific embodiment, it should be noted, however, that the specific implementation
Example does not constitute an undue limitation on the present application merely to the application is better described.
A kind of screenshot tool is provided in this example, and the automatic load to video frame or image is provided, a figure is being handled
Automatically next picture is loaded after piece.Meanwhile it also supporting to interrupt load, that is, the screenshot tool is closed after last time processing to half,
When next time opens, the processing of last time is continued directly to.When determining target area, it is only necessary to click several lower left buttons, a lower right
Key, program will automatically calculate target ROI, rename automatically, automatically save target ROI, and title after just renaming automatically,
Coordinate points information preservation is into txt text file.Further, perspective transform and small automatically can also be carried out to target ROI
Angle random direction is micro- to be transferred to obtain diversified sample.
As shown in Fig. 2, may include steps of:
S1: Auto load Images:
The video frame or image data catalogue captured by camera is loaded, the frame or image detected under the catalogue automatically is
It is no processed, it filters out untreated data and is successively loaded, and be shown in corresponding Image control.
S2: 90 ° of continuous rotations of image:
In view of due to shoot problem, some images show be not it is positive, therefore, a rotation can be added and pressed
Button, it is every to click once, 90 ° are rotated clockwise, thus just by image rotation.
S3: rotation image coordinate mapping:
After image rotation, can with according to the pixel coordinate of the calculated original image of image rotation process in rotation
The mapping of the pixel coordinate in image afterwards, while postrotational image is replaced into original image and is saved in original image preservation
Position, operations all later is all based on the rotation image.
S4: lock onto target:
Using trained template (such as: car plate detection template, pedestrian detection template etc. are trained using machine learning
Detection template) to every load image carry out target detection, when mouse is moved to detected target, the target position
It will do it the automatic amplification of hypergeometric example mapping;For undetected target, mouse can be placed on to the approximate centre position of target
And key amplifies in a mouse click.
S5: target scale:
Amplify to target is conveniently handled it to show the target in image clearly.By
Not only one target in image, it is possible to by shrinking of object to original image size, find other targets and carry out same
The amplification process of sample (that is, after the complete target of enhanced processing, by image down to original size, finds another target and again
Secondary enhanced processing, until all targets have all been handled).
When amplification, 9 pieces can be divided the image into: the upper left corner (0), the upper right corner (1), the lower left corner (2), the lower right corner
(3), upper (4) in, under (5), left (6), (7), intermediate (8) in the right side, due to the size of rectangle frame in control be it is fixed, because
The frame, can be named as PICTURE_1 and be shown on PICTURE_1 after picture load by this, realize that amplification can only be part
Amplification, that is, local ROI is replaced into original image out from interception in original image and is shown in PICTURE_1, next time puts again
When big, a part ROI in the image of last time amplification is intercepted out, the image for replacing last time is shown to PICTURE_ again
In 1, and so on;During entire display, control size is constant, and image-region range is smaller and smaller, so part can be put
Big display;For 9 blocks, difference is realized in each piece of amplification, and 1/4,1 piece of the upper left corner amplification of 0 piece of amplification present image is worked as
It puts in 1/4,4 piece of the lower right corner of 1/4, the 3 piece of amplification present image in the lower left corner of 1/4, the 2 piece of amplification present image in the upper right corner of preceding image
For big width centered on mouse position, the 1/4 of present image width, the highly upper half for present image height are respectively expanded in both sides
The partially image at (1/2), i.e., upper 1/4,5 piece of amplification width is centered on mouse position in amplification present image, and two
The 1/4 of present image width is respectively expanded on side, is highly the image at the lower half portion (1/2) of present image height, i.e. amplification is current
Lower 1/4,6 piece of amplification height respectively expands the 1/4 of present image height, width centered on mouse position up and down in image
For the image at the left-half (1/2) of present image width, i.e., left 1/4,7 piece of amplification height of amplification present image is with mouse
Centered on marking position, respectively expand the 1/4 of present image height up and down, width is the right half part (1/2) of present image width
The image at place, i.e., 1/4,8 piece of amplification height is each up and down to expand currently centered on mouse position in the right side of amplification present image
The 1/4 of picture altitude, for width centered on mouse position, the 1/4 of present image width is respectively expanded on both sides, i.e. the current figure of amplification
The middle section 1/4 of picture.It reduces and realizes: reducing and realize to be exactly to amplify the inverse process realized, every diminution successively, will will currently be schemed
It is shown in PICTURE_1 as that image before amplification is substituted into present image, until returning to most original image size.Amplification
Mode is by key in a mouse click, then amplification is to intercept the 1/4 of present image centered on mouse position every time, and amplification is aobvious
Show in big PICTURE_1 as present image, key (idler wheel) is pressed amplification process and successively reduced in sliding mouse, until former
Beginning image size.
S6: target ROI is obtained:
In the target of amplification, selects key point and click left mouse button, several key point dots can be drawn in the picture
And the coordinate put, after clicking right mouse button, boundary rectangle, on demand the ratio of width to height expansion are successively acquired according to key point coordinate
Rectangle afterwards, final rectangle are target ROI.
S7: file renaming and the storage of ROI information:
Target ROI image or original image are renamed according to demand, the pass of title and correspondence image will be renamed
Key point (coordinate points that left mouse button is clicked) coordinate record is stored in line by line in txt text file.
S8: sample is generated:
Target ROI image is saved hereof, the sample as generation.
S9: sample diversification:
There are some samples (such as: license plate sample) due to artificial shooting angle problem, the sample for directly taking its ROI to obtain is simultaneously
It is unavailable, the combination of perspective transform and the fine tuning of low-angle random direction is carried out to the key point of these targets, then carries out width again
Height can be obtained by available and diversified sample than expanding in this way.
By mode provided by upper example, ten pictures can be handled within one minute, relative to existing processing side
The case where being only capable of one two kinds of processing for one minute in formula, the application treatment effeciency is obviously some higher.
Based on the same inventive concept, a kind of target object selecting device is additionally provided in the embodiment of the present invention, it is such as following
Described in embodiment.Since the principle that target object selecting device solves the problems, such as is similar to target object choosing method, target
The implementation of object select device may refer to the implementation of target object choosing method, and overlaps will not be repeated.It is following to be used
, the combination of the software and/or hardware of predetermined function may be implemented in term " unit " or " module ".Although following embodiment institute
The device of description preferably realized with software, but the combined realization of hardware or software and hardware be also may and quilt
Conception.Fig. 3 is a kind of structural block diagram of the target object selecting device of the embodiment of the present invention, as shown in figure 3, may include:
Loading module 301, detection module 302, amplification module 303, display module 304 and logging modle 305 below carry out the structure
Explanation.
Loading module 301 is used for Auto load Images;
Detection module 302, for carrying out target detection to the image of load by template trained in advance;
Amplification module 303, in the case where detecting that mouse is moved to the target area detected, to being examined in image
The target area measured carries out the automatic amplification of hypergeometric example mapping, there is the case where target area being not detected in the picture
Under, in response to the clicking operation to the target area, the mapping of hypergeometric example is carried out to the target area being not detected and is put automatically
Greatly;
Display module 304 shows the key point in the target area for mapping automatic amplified region in hypergeometric example,
And the boundary rectangle of each key point is sought, using boundary rectangle as target object;
Logging modle 305 for the target object to be renamed and recorded, and records the target object and is adding
Carry the coordinate in image.
In one embodiment, above-mentioned target object selecting device can also include: conversion module, for described
Target object is renamed and is recorded, and after recording coordinate of the target object in load image, to the target
Object carries out perspective transform and the fine tuning of low-angle random direction, obtains the image for converting target object adjusted;Transformation is adjusted
The image of target object after whole carries out the ratio of width to height expansion.
In an implementation method, amplification module 303 may include: the first determination unit, for determining that clicking operation exists
Corresponding click location in image;Second determination unit, for determining magnification region according to click location;Display unit is used for
The position that magnification region is replaced to image, is shown in control, wherein the size of control is fixed.
In one embodiment, above-mentioned second determination unit may include: division subelement, for dividing an image into
Nine regions, wherein nine regions include: the upper left corner 1/4 of present image, it is the upper right corner 1/4 of present image, current
The lower right corner 1/4 of image, the lower left corner 1/4 of present image, in present image lower 1/4 in upper 1/4, present image, it is current
Left the 1/4 of image, in the right side of present image 1/4, present image centre 1/4;Subelement is selected, is used for from nine areas
Select a region as region to be amplified in domain.
In another embodiment, a kind of software is additionally provided, the software is for executing above-described embodiment and preferred reality
Apply technical solution described in mode.
In another embodiment, a kind of storage medium is additionally provided, above-mentioned software is stored in the storage medium, it should
Storage medium includes but is not limited to: CD, floppy disk, hard disk, scratch pad memory etc..
It can be seen from the above description that the embodiment of the present invention realizes following technical effect: in the embodiment of the present invention
In, a kind of target object choosing method and device are provided, can be put automatically with Auto load Images, automatic detected target object
Big target object, and the position of key point is automatically recorded, target object is renamed and is automatically stored, to solve existing
Target object choose the excessively cumbersome technical problem of process, reached the technology effect for being simple and efficient and realizing that target object is chosen
Fruit.
Obviously, those skilled in the art should be understood that each module of the above-mentioned embodiment of the present invention or each step can be with
It is realized with general computing device, they can be concentrated on a single computing device, or be distributed in multiple computing devices
On composed network, optionally, they can be realized with the program code that computing device can perform, it is thus possible to by it
Store and be performed by computing device in the storage device, and in some cases, can be held with the sequence for being different from herein
The shown or described step of row, perhaps they are fabricated to each integrated circuit modules or will be multiple in them
Module or step are fabricated to single integrated circuit module to realize.In this way, the embodiment of the present invention be not limited to it is any specific hard
Part and software combine.
The foregoing is only a preferred embodiment of the present invention, is not intended to restrict the invention, for the skill of this field
For art personnel, the embodiment of the present invention can have various modifications and variations.All within the spirits and principles of the present invention, made
Any modification, equivalent substitution, improvement and etc. should all be included in the protection scope of the present invention.
Claims (10)
1. a kind of target object choosing method characterized by comprising
Auto load Images;
By template trained in advance, target detection is carried out to the image of load;
In the case where detecting that mouse is moved to the target area detected, the target area detected in image is surpassed
The automatic amplification of ratio mapping, in the case where there is the target area being not detected in the picture, in response to the target area
Clicking operation, the automatic amplification of hypergeometric example mapping is carried out to the target area being not detected;
Automatic amplified region is mapped in hypergeometric example and shows the key point in the target area, and seeks the external of each key point
Rectangle, using boundary rectangle as target object;
The target object is renamed and recorded, and records coordinate of the target object in load image.
2. the method according to claim 1, wherein the target object is renamed and is recorded, and
After recording coordinate of the target object in load image, the method also includes:
Perspective transform and the fine tuning of low-angle random direction are carried out to the target object, obtain converting target object adjusted
Image;
The ratio of width to height expansion is carried out to the image for converting target object adjusted.
3. the method according to claim 1, wherein after Auto load Images, the method also includes:
Determine whether the image of load needs to rotate;
In the case where determining that needs rotate, successively the image clockwise of load is rotated by 90 °, until just by image rotation;
The original image for rotating positive image and replacing load is stored.
4. the method according to claim 1, wherein in response to carrying out clicking operation to the target area, to this
The target area being not detected carries out the automatic amplification of hypergeometric example mapping, comprising:
Determine clicking operation corresponding click location in the picture;
According to click location, magnification region is determined;
The position that magnification region is replaced to image, is shown in control, wherein the size of control is fixed.
5. according to the method described in claim 4, it is characterized in that, determining magnification region according to click location, comprising:
Divide an image into nine regions, wherein nine regions include: the upper left corner 1/4 of present image, present image
The upper right corner 1/4, the lower right corner 1/4 of present image, the upper right corner 1/4 of present image, upper 1/4 in present image, current figure
Lower 1/4 in picture, left the 1/4 of present image, in the right side of present image 1/4, present image centre 1/4;
Select a region as region to be amplified from nine regions.
6. the method according to any one of claims 1 to 5, which is characterized in that Auto load Images, comprising:
Load the catalogue of the video frame or image data that capture by camera;
Automatically it detects frame under the catalogue or whether image is processed;
If untreated, successively untreated frame or image are loaded onto Image control.
7. a kind of target object selecting device characterized by comprising
Loading module is used for Auto load Images;
Detection module, for carrying out target detection to the image of load by template trained in advance;
Amplification module, in the case where detecting that mouse is moved to the target area detected, to what is detected in image
Target area carries out the automatic amplification of hypergeometric example mapping, in the case where there is the target area being not detected in the picture, response
In the clicking operation to the target area, the automatic amplification of hypergeometric example mapping is carried out to the target area being not detected;
Display module shows the key point in the target area for mapping automatic amplified region in hypergeometric example, and seeks
The boundary rectangle of each key point, using boundary rectangle as target object;
Logging modle for the target object to be renamed and recorded, and records the target object in load image
In coordinate.
8. device according to claim 7, which is characterized in that further include:
Conversion module for the target object to be renamed and recorded, and records the target object in loading figure
After coordinate as in, perspective transform is carried out to the target object and low-angle random direction is finely tuned, after obtaining transformation adjustment
Target object image;The ratio of width to height expansion is carried out to the image for converting target object adjusted.
9. device according to claim 7, which is characterized in that the amplification module includes:
First determination unit, for determining clicking operation corresponding click location in the picture;
Second determination unit, for determining magnification region according to click location;
Display unit is shown in control for magnification region to be replaced to the position of image, wherein the size of control is
Fixed.
10. device according to claim 9, which is characterized in that second determination unit includes:
Subelement is divided, for dividing an image into nine regions, wherein nine regions include: the upper left of present image
Angle 1/4, the upper right corner 1/4 of present image, the lower right corner 1/4 of present image, the lower left corner 1/4 of present image, present image
In upper 1/4, lower 1/4 in present image, left the 1/4 of present image, in the right side of present image 1/4, present image centre
1/4;
Subelement is selected, for selecting a region as region to be amplified from nine regions.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710427979.4A CN109034173A (en) | 2017-06-08 | 2017-06-08 | Target object choosing method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710427979.4A CN109034173A (en) | 2017-06-08 | 2017-06-08 | Target object choosing method and device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109034173A true CN109034173A (en) | 2018-12-18 |
Family
ID=64629502
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710427979.4A Pending CN109034173A (en) | 2017-06-08 | 2017-06-08 | Target object choosing method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109034173A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110110148A (en) * | 2019-05-28 | 2019-08-09 | 成都德迈安科技有限公司 | A kind of commodity search method for video image |
CN112035038A (en) * | 2020-08-31 | 2020-12-04 | 北京字节跳动网络技术有限公司 | Picture processing method, device, equipment and storage medium |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101375800A (en) * | 2007-08-31 | 2009-03-04 | 深圳市蓝韵实业有限公司 | Method for amplifying and displaying ultrasonoscopy |
CN102982350A (en) * | 2012-11-13 | 2013-03-20 | 上海交通大学 | Station caption detection method based on color and gradient histograms |
CN103345738A (en) * | 2013-06-04 | 2013-10-09 | 北京百度网讯科技有限公司 | Object detection method and device based on area of interest |
CN103645871A (en) * | 2013-12-06 | 2014-03-19 | 四川九洲电器集团有限责任公司 | Method and system for dynamically displaying local area image |
CN103942577A (en) * | 2014-04-29 | 2014-07-23 | 上海复控华龙微***技术有限公司 | Identity identification method based on self-established sample library and composite characters in video monitoring |
CN105005757A (en) * | 2015-03-12 | 2015-10-28 | 电子科技大学 | Method for recognizing license plate characters based on Grassmann manifold |
CN106780525A (en) * | 2016-11-18 | 2017-05-31 | 北方工业大学 | Optical remote sensing image ship direction feature extraction method based on coordinate rotation minimum circumscribed rectangle |
-
2017
- 2017-06-08 CN CN201710427979.4A patent/CN109034173A/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101375800A (en) * | 2007-08-31 | 2009-03-04 | 深圳市蓝韵实业有限公司 | Method for amplifying and displaying ultrasonoscopy |
CN102982350A (en) * | 2012-11-13 | 2013-03-20 | 上海交通大学 | Station caption detection method based on color and gradient histograms |
CN103345738A (en) * | 2013-06-04 | 2013-10-09 | 北京百度网讯科技有限公司 | Object detection method and device based on area of interest |
CN103645871A (en) * | 2013-12-06 | 2014-03-19 | 四川九洲电器集团有限责任公司 | Method and system for dynamically displaying local area image |
CN103942577A (en) * | 2014-04-29 | 2014-07-23 | 上海复控华龙微***技术有限公司 | Identity identification method based on self-established sample library and composite characters in video monitoring |
CN105005757A (en) * | 2015-03-12 | 2015-10-28 | 电子科技大学 | Method for recognizing license plate characters based on Grassmann manifold |
CN106780525A (en) * | 2016-11-18 | 2017-05-31 | 北方工业大学 | Optical remote sensing image ship direction feature extraction method based on coordinate rotation minimum circumscribed rectangle |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110110148A (en) * | 2019-05-28 | 2019-08-09 | 成都德迈安科技有限公司 | A kind of commodity search method for video image |
CN112035038A (en) * | 2020-08-31 | 2020-12-04 | 北京字节跳动网络技术有限公司 | Picture processing method, device, equipment and storage medium |
CN112035038B (en) * | 2020-08-31 | 2022-03-04 | 北京字节跳动网络技术有限公司 | Picture processing method, device, equipment and storage medium |
EP4191391A4 (en) * | 2020-08-31 | 2023-09-13 | Beijing Bytedance Network Technology Co., Ltd. | Picture processing method and apparatus, device, and storage medium |
US11941777B2 (en) | 2020-08-31 | 2024-03-26 | Beijing Bytedance Network Technology Co., Ltd. | Picture enlargement method and apparatus, device, and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Ye et al. | Robust registration of multimodal remote sensing images based on structural similarity | |
US5592573A (en) | Method and apparatus for determining mis-registration | |
Sharma et al. | Image stitching using AKAZE features | |
KR20100112155A (en) | Feature-based signatures for image identification | |
CN105893405A (en) | Repetitive video detection method and system | |
Shah et al. | Removal of specular reflections from image sequences using feature correspondences | |
US10152646B2 (en) | Keypoint identification | |
CN112001219A (en) | Multi-angle multi-face recognition attendance checking method and system | |
CN109034173A (en) | Target object choosing method and device | |
WO2022267939A1 (en) | Image processing method and apparatus, and computer-readable storage medium | |
CN112052702A (en) | Method and device for identifying two-dimensional code | |
US7113652B2 (en) | System and method for using normalized gray scale pattern find | |
Brogan et al. | Spotting the difference: Context retrieval and analysis for improved forgery detection and localization | |
CN113689458B (en) | 2D shooting track path calculation method and device | |
US8705874B2 (en) | Image processing method and system using regionalized architecture | |
US20030185431A1 (en) | Method and system for golden template image extraction | |
JPS62107386A (en) | Image matching method | |
Elaskily et al. | Enhanced Fiilltterr-based SIFT Apprroach fforr Copy-Move Forrgerry Dettecttiion | |
CN111768338A (en) | Method and device for splicing test question images, electronic equipment and storage medium | |
Kasar et al. | Block-based feature detection and matching for mosaicing of camera-captured document images | |
Andrade et al. | A robust methodology for outdoor optical mark recognition | |
Chu et al. | An information theoretic framework for order of operations forensics | |
JP2005141458A (en) | Image processing method and device, program, and recording medium for the program | |
CN111160486B (en) | Fuzzy image classification method based on support vector machine and wavelet decomposition | |
JP3127089B2 (en) | Image processing apparatus and image processing method |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181218 |