CN107018407A - Information processor, evaluation figure, evaluation system and method for evaluating performance - Google Patents
Information processor, evaluation figure, evaluation system and method for evaluating performance Download PDFInfo
- Publication number
- CN107018407A CN107018407A CN201710025681.0A CN201710025681A CN107018407A CN 107018407 A CN107018407 A CN 107018407A CN 201710025681 A CN201710025681 A CN 201710025681A CN 107018407 A CN107018407 A CN 107018407A
- Authority
- CN
- China
- Prior art keywords
- image
- characteristic point
- evaluation
- extracted
- performance
- 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N17/00—Diagnosis, testing or measuring for television systems or their details
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T1/00—General purpose image data processing
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N17/00—Diagnosis, testing or measuring for television systems or their details
- H04N17/002—Diagnosis, testing or measuring for television systems or their details for television cameras
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
- Studio Devices (AREA)
- Image Processing (AREA)
Abstract
The present invention relates to information processor, evaluate with figure, evaluation system and method for evaluating performance.Efficiently find out the condition for the information processing for for high accuracy used shooting image.Camera device (12) is shot evaluates with figure (200), and same method generates captured image data and sent to information processor (10) during with application.Information processor (10) is being evaluated with figure (200), detects the corner of the big rectangle of checkerboard, by the relative position with it, obtains the coordinate in corner of the small rectangle of the grey internally represented as normal solution coordinate.Same algorithm is to extract characteristic point when also with application, and is compared with normal solution coordinate, so as to calculate the index for representing the performance in the viewpoint of feature point extraction.In the case where the index is not reaching to desired value, information processor (10) updates and adjusts the setting of regulating object among the setting of the various processing in camera device (12).
Description
Technical field
The present invention relates to the information processing system for the information processing for carrying out the identification comprising the real object for having used shooting image
System, the camera device wherein included, pad (mat) and used image generating method.
Background technology
It is known that the body of user, mark are shot by camera, by the regional replacement of the picture be other images simultaneously
It is shown in the game of display (for example, referring to patent document 1).By detecting and analyzing the picture in shooting image, obtain shot
The position of body, camera itself, motion or identification subject are that what technology make use of by extensive importing and be equipped on playsuit
Put, the camera of information terminal and make use of in the information processing that security camera, in-vehicle camera, robot carry camera etc..
In technology as described above, it is important all the time that the picture of object, the feature of its appearance how are obtained exactly
Problem.Therefore the profile of the object extracted exactly in shooting image, the various technologies at the position of feature are proposed (for example, ginseng
According to patent document 2).
Prior art literature
Patent document
Patent document 1：No. 0999518 specification of European Patent Application Publication
Patent document 2：(Japan) JP 2010-182078 publications
The content of the invention
The position-based of feature in shooting image is in equipment used in the shooting environmentals such as lightness, imaging sensor etc.
Performance, be applied in untill feature detection from shooting data various processing, specifically solution mosaic processing, at correction
Condition, method in reason, compressed encoding processing, communication process etc., it extracts easness and changed.It is used as so multiple main causes
The result intricately associated, due to used the information processing of detected feature, so being not easy to find out for full
The optimum condition for restricting and obtaining sufficient processing accuracy in the manufacturing cost of pedal system, the viewpoint of processing cost.
The present invention be in view of such problem and complete, its object is to there is provided can efficiently find out for high-precision
Degree ground used the technology of the treatment conditions of the information processing of shooting image.
In order to solve above-mentioned problem, certain mode of the invention is related to information processor.The feature of the information processor
It is possess：Figure image acquiring section, the data of the shooting image of the evaluation figure of pattern as defined in representing are obtained from camera device
And deploy in memory as image；Feature point extraction portion, characteristic point is extracted from shooting image；Normal solution coordinate obtaining section,
In the method different from feature point extraction portion, the position coordinates for obtaining the characteristic point that should be extracted from shooting image is sat as normal solution
Mark；And performance Index Calculation portion, characteristic point and normal solution coordinate that distinguished point based extraction unit is extracted, calculating will be suitable to
Performance indications after the performance number value of the image of feature point extraction are simultaneously exported.
The other modes of the present invention are related to evaluation figure.The evaluation with figure is suitable to for evaluation information processing unit
The evaluation figure of the performance of the image of feature point extraction, it is characterised in that include the chess for detecting corner from the shooting image
The black and white rectangle of plate-like and each rectangle internal representation grey figure, should each rectangle internal representation
The figure of grey is handled with the feature point extraction for being obtained coordinate based on the testing result in corner and with being additionally carried out
The characteristic point of results contrast.
The other modes again of the present invention are directed to evaluation figure.The evaluation with figure is obtained for evaluation information processing unit
Evaluation figure suitable for the performance of the image of feature point extraction, it is characterised in that include being configured to for being detected from the shooting image
In the figure for the grey for forming multiple black circles in the space of an amount up and down and being represented in the inside of each circle and the space
Shape, the figure for the grey that should be represented in the inside of each circle and the space, which has, to be used to obtain based on the testing result of the circle
Coordinate and be additionally carried out feature point extraction processing results contrast characteristic point.
The other modes again of the present invention are related to evaluation system.The evaluation system is characterised by, regulation is represented comprising shooting
Pattern evaluation figure camera device and obtain the data of the shooting image and evaluation obtains figure suitable for feature point extraction
The information processor of the performance of picture, information processor possesses：Figure image acquiring section, shooting image is obtained from camera device
Data are simultaneously deployed in memory as image；Feature point extraction portion, characteristic point is extracted from shooting image；Normal solution coordinate is obtained
Portion, in the method different from feature point extraction portion, the position coordinates for obtaining the characteristic point that should be extracted from shooting image is used as normal solution
Coordinate；Performance Index Calculation portion, characteristic point and normal solution coordinate that distinguished point based extraction unit is extracted, is calculated after performance number value
Performance indications；And adjustment portion, when performance indications are not reaching to desired value set in advance, to the processing in camera device
One of them of used setting is adjusted, and camera device possesses：Parameter storage part, storage is adjusted by adjustment portion
Setting；And data forming portion, according to the setting that parameter storage part is stored, the Data Concurrent that generates shooting image delivers to letter
Cease processing unit.
The other modes again of the present invention are related to method of evaluating performance.The method of evaluating performance is characterised by, comprising：From taking the photograph
As device obtain represent as defined in pattern evaluation figure shooting image data and deploy in memory as image
The step of；The step of characteristic point being extracted from shooting image；The step of with from extracting characteristic point different method, acquirement should be from shooting
The step of position coordinates of the characteristic point of image zooming-out is as normal solution coordinate；And based on the extraction in the step of extracting characteristic point
The characteristic point and normal solution coordinate arrived, calculate by obtain suitable for feature point extraction image performance number value after performance indications simultaneously
The step of being exported.
In addition, by the arbitrary combination of structural element above, method, device, system, record are situated between for of the invention showing
Between matter, computer program etc. change after combination, performance as the present invention mode be also effective.
According to the present invention, the condition for accurately carrying out having used the information processing of shooting image is efficiently found out.
Brief description of the drawings
Fig. 1 is to represent that the figure of the configuration example of the evaluation system of present embodiment can be applied.
Fig. 2 is the figure for the internal circuit configuration for representing the information processor in present embodiment.
Fig. 3 is the figure for the main cause for illustrating the object for being envisioned for adjustment in the present embodiment.
The structure of the functional block of information processor and camera device when Fig. 4 is the evaluation for representing present embodiment
Figure.
Fig. 5 is the figure for illustrating the evaluation figure in present embodiment.
Fig. 6 (a)~(b) is for illustrating the normal solution coordinate obtaining section in present embodiment from the bat of Fig. 5 evaluation figure
Take the photograph image and obtain the figure that normal solution sits calibration method.
Fig. 7 (a)~(b) be in the present embodiment by normal solution coordinate obtaining section achieve the small rectangle of coordinate corner,
The figure that the characteristic point extracted with feature point extraction portion is compared.
Fig. 8 is to illustrate in the present embodiment, the distribution of the characteristic point of the image zooming-out to have taken evaluation figure and institute
The figure of the F values calculated.
Fig. 9 is to illustrate in the present embodiment, the distribution of the characteristic point of the image zooming-out to have taken evaluation figure and institute
The figure of the F values of calculating.
Figure 10 is to illustrate in the present embodiment, the distribution of the characteristic point of the image zooming-out to have taken evaluation figure and
The figure of the F values calculated.
Figure 11 is to illustrate in the present embodiment, the distribution of the characteristic point of the image zooming-out to have taken evaluation figure and
The figure of the F values calculated.
Figure 12 be illustrate S/N than figure of the change relative to the change of unit size.
Figure 13 is to represent in the present embodiment, the data knot of the set information of the desired value stored in desired value storage part
The figure of structure example.
Figure 14 is to represent that information processor is exchanged on the basis of have rated feature point extraction performance in the present embodiment
The flow chart for the processing procedure that whole image parameter is adjusted.
Figure 15 is the figure of another for representing the evaluation figure in present embodiment.
Label declaration
8 evaluation systems, 10 information processors, 12 camera devices, 16 display devices, 22CPU, 24GPU, 26 primary storages
Device, 60 data forming portions, 62 image pickup parts, 64 parameter storage parts, 70 figure image acquiring sections, 71 image storage portions, 72 characteristic points are carried
Take portion, 74 normal solution coordinate obtaining sections, 76 performance Index Calculation portions, 78 comparing sections, 80 desired value storage parts, 82 adjustment portions, 84 ginsengs
Number candidate's storage part, 200 evaluations figure, 300 evaluations figure.
Embodiment
Present embodiment is directed to the information processing for accurately carrying out having used the characteristic point extracted from shooting image
And the performance evaluation carried out in advance and adjustment as needed.The performance evaluation, adjustment can also be at exploitation, manufacture information
The stage of reason system is carried out by manufacturer etc., can also be carried out by the user for having bought product in actual use environment.No
It is particularly limited to the content using information processings such as the game of characteristic point progress.After, certain will be carried out using the characteristic point extracted
The stage of a little information processings, which is referred to as " when applying ", will carry out performance evaluation, the stage of adjustment before is referred to as " during evaluation ".
Fig. 1 represents that the configuration example of the evaluation system of present embodiment can be applied.Evaluation system 8 is included to be schemed to evaluation
At 200 camera devices 12 shot, the information evaluated from taken image zooming-out characteristic point and to its result
Reason device 10 and it regard the information such as evaluation result as image and the display device 16 that exports.In addition, with the system architecture shown in figure
It is system architecture when evaluating, in use, can also be that 12 pairs of camera device space shot corresponding with purposes is shot,
Information processor 10 carries out the information processing of user's request etc..In addition, display device 16 can also show the information processing
As a result.
Camera device 12 has the camera shot to object, implements solution mosaic processing etc. one to its output signal
As processing so as to generate shooting image output data and to information processor 10 send out mechanism.Camera possesses CCD (electricity
Lotus coupled apparatus (Charge Coupled Device)) sensor, CMOS (complementary metal oxide semiconductors
(Complementary Metal Oxide Semiconductor)) sensor etc. is in general digital camera, DV
The middle visible light sensor utilized.The camera that camera device 12 possesses can also only one of which or by two cameras with
Known interval is configured at the so-called stereoscopic camera of left and right.
Information processor 10 for example can also be game device, personal computer, be evaluated, using required by loading
Program realizes the information processing function.Or information processor 10 can also be set to be directed to performance evaluation, the device of adjustment.
In this case, exchanged in application as devices corresponding with purposes such as game device, personal computers.
Display device 16 can also be the general display such as liquid crystal display, plasma display, organic el display.
The information such as the result for the evaluation that display device 16 display information processor 10 is carried out, characteristic point for being extracted when evaluating.
, can also be without display device 16 when evaluating in the case of the prompting for not needing such information.
Camera device 12, information processor 10, the connection of display device 16 can also be it is wired can also be wireless.
The image shot in present embodiment in particular according to camera device 12, evaluates whether to be extracted characteristic point with sufficient precision.
And as needed, to the processing inside camera device 12, from camera device 12 to the transmission of the data of information processor 10, spy
The condition for levying extraction process a little etc. is adjusted.It is purpose due to so determining appropriate condition and applying flexibly in application, so
Camera device 12, communication component follow the structure contemplated in application.
Fig. 2 represents the internal circuit configuration of information processor 10.(center processing is single comprising CPU for information processor 10
First (Central Processing Unit)) 22, GPU (graphics processing unit (Graphics Processing Unit)) 24,
Main storage 26.CPU22 based on operating system, using etc. program, in the structural element inside information processor 10
Processing, signal transmission are controlled.GPU24 carries out image procossing.Main storage 26 is by RAM (random access memory (Random
Access Memory)) constitute, program, data needed for storage processing.
These each portions are connected with each other via bus 30.Input/output interface 28 is also associated with bus 30.In input
On output interface 28, it is connected with what is be made up of the peripheral equipment such as USB, IEEE1394 interface, wired or wireless LAN network interface
Communication unit 32, the storage part 34 such as hard drive, nonvolatile memory is exported to output devices such as display device 16, loudspeakers
The output section 36 of data, it is removable to disk, CD or semiconductor memory etc. from the input unit 38 of the input data of camera device 12
Unload the recording medium drive division 40 that recording medium is driven.
CPU22 is by performing the operating system stored in storage part 34 so as to the entirety of control information processing unit 10.
CPU22 also performs the various journeys for being read from detachable recording medium and being loaded into main storage 26 or being downloaded via communication unit 32
Sequence.GPU24 has the function of geometry engines and the function of rendering processor, is described according to the drawing command from CPU22
Processing, display image is stored to frame buffer (not shown).And the display image stored in frame buffer is converted to and regarded
Frequency signal is simultaneously exported to output section 36.
When evaluating, information processor 10 is from the image zooming-out characteristic point of the figure that have taken evaluation, in method described later
Calculate and represent that the numerical value of its performance is used as performance indications.And the display of display device 16 is extracted result, performance indications, or be based on
It adjusts shooting condition, treatment conditions.But can also be the display for only carrying out result, it is seen that its developer, user passes through
Human hand and change condition, without all of which is implemented, can also only adjust automatically treatment conditions and without information
Display.Under any circumstance, by being adjusted repeatedly untill good performance is obtained, so as to finally give in characteristic point
The condition of optimal image is obtained in the viewpoint of extraction.
In this as regulating object " condition " can also be from shooting to feature point extraction processing untill during meeting
One of them or one of main cause influential on extraction accuracy can also be multiple combinations.Fig. 3 is illustrated and set
Want the main cause of the object for adjustment.As shown in the figure, it is believed that the species for the imaging sensor that camera device 12 has, shooting bar
Part, the algorithm used when generating view data according to the output valve of sensor, the wave filter used, carry out to view data
The various main causes pair such as mode, setting value during compressed encoding, wireless mode when sending to information processor 10 data
The extraction accuracy of shooting image and then characteristic point has an impact.
The key element of illustration is pursued on the direction that people suitably recognized after seeing when showing image in the past.But so
The condition of decision is not necessarily favourable to feature point extraction.It is therefore contemplated that even with from the shooting figure obtained with such condition
As the characteristic point extracted carries out information processing, the processing accuracy that will not be also met, or do not notice and be easy to estimate into one
The precision of step is improved.
Therefore, in the present embodiment, the stage untill the extraction to characteristic point, evaluated from the viewpoint of performance is extracted
Image, thus processing accuracy will be given influence the reason for separated with the information processing of rear class.In addition, by performance indications numerical value
Change, can readily be compared, so as to determine the big main cause of influence from the substantial amounts of main cause of diagram, or it is high
Imitate ground and then automatically determine optimal conditions.In addition, as the main cause of regulating object as illustrated, including image sensing
Species, algorithm of device etc. individually have the numerical value such as the main cause and time for exposure, gamma value of entity.In the following description will
They are blanket and are referred to as " regulating object parameter ".
The structure of the functional block of information processor 10 and camera device 12 when Fig. 4 represents to evaluate.It is each shown in Fig. 4
Functional block can be realized on hardware by CPU, GPU, main storage etc., by being loaded into master from hard disk, recording medium on software
Computer program of memory etc. is realized.So as to, these functional blocks can by only hardware, only software or combinations thereof with
Various forms are realized for a person skilled in the art it should be appreciated that be not limited to one of them.
Camera device 12 should be also sent to information in addition to the image pickup part 62 being made up of imaging sensor comprising formation
Manage the data forming portion 60 of the data of the shooting image of device 10, store the setting of the parameter formed for shooting condition, data
The parameter storage part 64 of value.In the parameter, regulating object parameter is included.Image pickup part 62 includes the image sensings such as CCD, CMOS
Device, shoots evaluation figure.Now, with according to setting values such as time for exposure, stop value, the yield values stored in parameter storage part 64
Condition imaged.
Data forming portion 60 generates the number of shooting image according to the electric signal by each pixel exported from image pickup part 62
According to.Specifically, solution mosaic processing, the correction process of defect pixel, gamma correction etc. are carried out.In addition, data forming portion 60
The data of image to being generated are compressed coding, and information processor 10 is sent to defined communication protocol.On
Algorithm, wave filter for these various processing etc., also according to the setting stored in parameter storage part 64.
Figure image acquiring section 70 of the information processor 10 comprising the data for obtaining the image that have taken evaluation figure, storage
The image storage portion 71 of the data of shooting image, with application when identical method from shooting image extract characteristic point characteristic point
Extraction unit 72, be suitable for evaluation figure method obtain correct characteristic point coordinate normal solution coordinate obtaining section 74, calculate special
The ratio that desired value set in advance and performance indications levy a performance Index Calculation portion 76 for the performance indications extracted, be compared to
Compared with portion 78, store the desired value storage part 80 of the desired value, based on comparative result to the adjustment that is adjusted of adjustment image parameter
The parameter candidate storage part 84 of the setting candidate of portion 82 and storage regulating object parameter.
Figure image acquiring section 70 obtains the data for the image that have taken figure from camera device 12, is compiled in accordance with the compression used
Code form is decoded, extended and deployed in image storage portion 71.Feature point extraction portion 72 is read from image storage portion 71 to scheme
The data of picture, identical algorithm extracts characteristic point during with application.Characteristic point is extracted as with purposes such as image recognitions
Technology, Harris, Eigen, Fast (Fast Segmentation test feature (Features from Accelerated Segment
Test)), SIFT (scale invariant feature changes (Scale-Invariant Feature Transform)), SURF (quick Shandongs
Rod feature (Speeded Up Robust Features)) etc. various algorithms be practical.
In the present embodiment using whichever will do, suitably selected according to purpose of information processing etc..Or can also
It is enough that algorithm is set to regulating object parameter in itself.Each algorithm is general algorithm so in this description will be omitted.
Normal solution coordinate obtaining section 74 reads the data of the shooting image of figure from image storage portion 71, schemes using suitable for evaluation
Defined algorithm, obtain the characteristic point that should extract originally on the image plane position coordinates (after, be referred to as that " normal solution is sat
Mark ").That is, to be extracted from feature point extraction portion 72, characteristic point is different to be directed to evaluation normal solution coordinate is obtained with the method for figure.
Concrete example is as be described hereinafter.
Performance Index Calculation portion 76 is by from the identical shooting image of evaluation figure, the spy extracted by feature point extraction portion 72
The normal solution coordinate levied point and obtained by normal solution coordinate obtaining section 74 is compareed, so as to regard the extraction performance of characteristic point as number
Value is represented.Performance Index Calculation portion 76 and then can also generate and with figure etc. represent feature point extraction portion 72 in shooting image
The image for the normal solution coordinate that the characteristic point and normal solution coordinate obtaining section 74 of extraction are obtained, and it is shown in display device 16.Thus,
Developer, user any degree can be proceeded to exactly in the extraction for sensuously understanding characteristic point.
Comparing section 78 reads the desired value of performance indications set in advance from desired value storage part 80, with performance Index Calculation
The actual performance indications that portion 76 is calculated are compared.On the desired value of performance indications, extracted characteristic point is used
The result of information processing is carried out, the value for ensureing to obtain sufficient processing accuracy is tried to achieve by experiment etc..Desired value can also pass through
The content of information processing, the subject for detecting object etc. and set multiple.In this case, by the every of these key elements, obtain
Optimal regulating object parameter.
Whether adjustment portion 82 reaches desired value according to performance indications to determine a need for adjustment.Result of determination can also show
Show in display device 16.In addition, when being judged to needing adjustment, adjustment portion 82 determines the new setting of regulating object parameter.Such as
It is upper described, for adjustment image parameter, it is considered to which time for exposure, stop value, compression ratio etc. continuously make the adjustment of numerical value change
Image parameter, the regulating object parameter from the single entity selection such as algorithm, wave filter.
In parameter candidate storage part 84, on making the regulating object parameter of numerical value change store adjustable number range
And its rule involved by the method for change, on the selectable candidate of regulating object parameter storage selected from candidate and its choosing
Select rule.Adjustment portion 82 determines new setting according to set rule, updates the parameter storage part 64 in camera device 12
The setting of middle storage.In adjustment, the difference of actual performance indications and desired value can also be reflected.That is, difference is more big, makes
The adjustment amplitude of regulating object parameter is bigger.
In addition, the parameter adjusted by once evaluating can also be one or multiple combinations.It can also be out
Originator, user can specify its object in advance.In addition, the difficult parameter of the adjust automatically such as species on imaging sensor, energy
It is enough to show result of determination etc. in display device 16 to be set to the research material of developer etc..If camera device 12 is used with more
The photographed images of the new evaluation figure for shooting and sending of setting after new, are repeated performance evaluation, the circulation of adjustment, energy again
It is enough efficiently to determine optimal conditions.
Fig. 5 illustrates evaluation figure.Evaluation can also depict figure pattern on the surface of the objects such as plate, pad with Figure 200,
The figure pattern displaying of electronic data can will be made in display device 16.By shooting such from front with defined distance
Evaluation Figure 200, so as to obtain illustrating such pattern as shooting image.The evaluation of diagram is with Figure 200 by between adjacent
Multiple big rectangles (such as big rectangle 202) of the checkerboard of white black reversion and in the small of the grey shown in the inside of major rectangles
Rectangle (such as small rectangle 204) is constituted.
Expecting the grey density of the small rectangle in evaluation Figure 200 has gradation.In this embodiment, whether can will be small
The corner (corner) of rectangle is accurate as characteristic point and the viewpoint do not extracted lavishly on carry out commenting for feature point extraction performance
Valency.That is, by apply when subject picture it is doubtful so that small rectangle is represented and evaluates extraction performance.So as to the ash shown in small rectangle
The weight of color is more, more represents the diversity of the picture in actual shooting image, can correctly be evaluated.
Fig. 6 is for illustrating that the shooting image for the evaluation figure that normal solution coordinate obtaining section 74 is shown from Fig. 5 obtains normal solution
Sit the figure of calibration method.Shown in addition with figure (a), (b) selected evaluation with a part among Figure 200 picture.Normal solution coordinate takes
Portion 74 is obtained first as shown in white fork mark in (a) of figure, the corner of big rectangle is extracted as characteristic point, obtains its position seat
Mark.From the shooting image of the pattern figure of checkerboard extract the corner of rectangle as the method for characteristic point in camera calibrated it is extensive
Known (referring for example to (Japan) JP 2014-92460 publications), extraction algorithm is also accurately real using general algorithm
Apply.
It can such as utilize in the field of computer vision in the widely used openCV increased income
CornerSubPix functions (http：//docs.opencv.org/2.4/modules/imgproc/doc/feature_
detection.html).Normal solution coordinate obtaining section 74 uses the method this ensure that precision, obtains the big square in shooting image
The position coordinates in the corner of shape.And as shown in white circle in (b) of figure, carried based on the relative position with the corner of big rectangle
Take the corner of small rectangle.
Specifically, as figure right side shown in, the position coordinates (x0, y0) in a corner of the picture based on big rectangle and
Laterally, length Δ x, the Δ y on the side of longitudinal direction, try to achieve the position coordinates (x, y) in the corner of the picture of corresponding small rectangle.Original
It is small if big rectangle is reduced into the small rectangular arrangement of n% (n ＜ 100) size in the center of big rectangle by handle in evaluation figure
The position coordinates (x, y) in the corner of rectangle is expressed as below.
【Number 1】
The positive and negative symbol on the right switches according to the position relationship in big rectangle corner corresponding with small rectangle herein.Separately
It is outer to expect the image to have taken evaluation figure to more precisely try to achieve normal solution coordinate, implement to remove based on lens etc.
The correction process of distortion.In such correction process, the general method, such as of the pattern figure of checkerboard can will have been used
The evaluation figure of present embodiment is directly applied to by the above-mentioned openCV functions provided.
Like this, by the way that the pattern used in conventional camera calibrated to be contained in a part for evaluation figure, so as to
Calibration process beyond the enough extraction performance evaluation for reaching characteristic point simultaneously.In addition, by the way that at least one of small rectangle is set to
Gray card with adjustment exposure, white balance is homochromy, so as to can also carry out these adjustment simultaneously.
On the other hand, as described above, feature point extraction portion 72 from identical shooting image with application when identical algorithm
To extract the corner of small rectangle.That is, relative to algorithm of the normal solution coordinate obtaining section 74 using the pattern for being directed to checkerboard, feature
Even point extraction unit 72 uses arbitrary subject also to extract the algorithm of characteristic point.
The evaluation of illustration is envisaged to be able to extract the algorithm in corner therein with figure, so being shown in the inside of big rectangle smaller
Rectangle.Wherein, can also be according to the feature point extraction algorithm used in application, by the figure shown in the inside of big rectangle
Shape be set to beyond rectangle.As long as with known to the position relationship in the corner of big rectangle, it becomes possible to independently obtained just with figure
Coordinate is solved, so extraction performance can be evaluated in the same manner as rectangle.
What the corner and feature point extraction portion 72 that Fig. 7 achieves normal solution coordinate obtaining section 74 in the small rectangle of coordinate were extracted
Characteristic point is compared.(a), among the shooting image of the evaluation figure shown in (b), (a) is surrounded with quadrangle shows that normal solution is sat
The position in the corner that obtaining section 74 is obtained is marked, (b) surrounds the characteristic point for showing that feature point extraction portion 72 is extracted with quadrangle.
As shown in (a), normal solution coordinate obtaining section 74 can extract the corner of small rectangle exactly by above-mentioned method.Separately
On the one hand, feature point extraction portion 72 can not extract the corner of small rectangle, or extract the position beyond corner by mistake.In addition in (b),
The quadrangle for surrounding the characteristic point extracted by mistake is represented with denseer line.
If gradually becoming strong mode on a direction of evaluation figure with the contrast of big rectangle and small rectangle makes small rectangle
Grey density change, then be readily able to grasp contrast with by mistake extract frequency relation.Performance Index Calculation portion 76 is by spy
The characteristic point for levying the extraction of an extraction unit 72 is shown as shown in (b), so that user can be lost with visually confirming to miss extraction, extraction
The trend of leakage, carries out the research of countermeasure.
Performance Index Calculation portion 76 will extract performance in the viewpoint that such extraction by mistake, extraction are omitted and quantize, so that
Can easily carry out it is multiple under the conditions of results contrast, or the automatic adjustment for being adjusted image parameter.Specifically, it will be used for
The yardstick for representing the performance of information retrieval etc. is that F values (F-measure) are calculated as below.
【Number 2】
I.e. F values are Precision (suitable rate) and Recall (recall factor) harmonic average.(it is adapted in this Precision
Rate) be the characteristic point that feature point extraction portion 72 is extracted number N among normal solution number R ratio, Recall (recall factor) is
Feature point extraction portion 72 among the number C for the normal solution coordinate that the number for the normal solution that should be extracted, i.e. normal solution coordinate obtaining section 74 are obtained
The number R of characteristic point ratio is extracted as, is given below respectively.
【Number 3】
By using F values, in addition to " precision " under wrong how much such implications in the characteristic point extracted,
Be also represented by comprising extract omit number viewpoint blanket extraction performance.Fig. 8~11 are illustrated to shooting under various conditions
The distribution of the characteristic point of the image zooming-out of evaluation figure and the F values calculated.In the various figures, will by with four kinds not
With photographing element camera and the result that shoots is compared, A, B, C have unit size (red green for large, medium and small RGB
It is blue) sensor, D has RGBIR (infrared ray) sensor.
But, either which sensor, all extracts feature from the luminance picture that the signal according to RGB is ultimately generated
Point.Fig. 8 represents that the bright place of the comparison in 700 luxs is shot, and carries out the result in the case of common Gain tuning.At this
Sample it is good under the conditions of shooting in, unit size can be reduced, whether there is infrared ray sensor caused by influence and stably obtain good
Good F values.
Fig. 9 represents to shoot with the place of Fig. 8 identical lightnesses, without the result in the case of Gain tuning.If
It is compared with the situation that carries out Gain tuning, then whichever sensor F value all deteriorates, but the degree deteriorated is different,
The small C of unit size sensor and also detect that ultrared D deterioration sensor is notable.
Figure 10 represents to shoot in the place than Fig. 8,9 100 dark luxs, in the case of having carried out common Gain tuning
Result.In this case, F values further deteriorate in the small C of unit size sensor.In addition it is known big in unit size
A, B sensor in, there is robustness relative to the change of the illumination of the degree.
Figure 11 represents the place shooting in 10 luxs more darker than Figure 10, in the case of having carried out common Gain tuning
Result.The image wherein shot in the sensor for also detecting ultrared D is because noise is without carrying out normal feature
Point is extracted so being set to not shown.In this case, in sensor of the unit size for the B of middle degree, F values are also significantly disliked
Change.
I.e. in the case where contemplating such dark place, it is known that for it can stand the information processing for having used characteristic point,
It is sensor more than setting to need to use unit size.In addition, being carried below the image being not shown in B, C sensor
The characteristic point got be based on when the number of error detection added up from top has exceeded setting by later characteristic point from aobvious
Show that object removes the rule on such software.
Figure 12 illustrate S/N than change of the change relative to unit size.As illustrated, S/N ratios have relative to unit
The increase of size linearly increased characteristic.On the other hand, in the result of Fig. 8~11, change of the F values relative to unit size
Can not be said to be linear, further according to shooting environmental illumination and its relation significantly changes.For example under identical unit size,
By border of certain illumination, F values tempestuously deteriorate, and the illumination as the border is different according to unit size.
So as to which the deterioration of the performance of feature point extraction processing depends not only upon noise, various main causes are additionally considered that intricately
It is related.According to the method illustrated in present embodiment, by such performance number value is easily carried out it is various under the conditions of
Comparison.Performance is evaluated thus, it is possible to distinguish multiple main causes, so can efficiently and easily carry out determining optimal ginseng
Several combinations, or confirm whether the information of characteristic point is able to maintain that the precision used when tolerance is applied.
Figure 13 represents the data structure example of the set information of the desired value stored in desired value storage part 80.Desired value is set
Table 260, which is included, applies column 262 and target value column 264.User's selection in the application such as each game is used as using the setting of column 262
And carry out the identification information of the application of the unit of information processing.It is corresponding with each application and set characteristic point in target value column 264
Extract the desired value of performance indications.
According to application, there is a situation where the higher extraction performance of requirement, on the other hand, also exist and less pursue characteristic point
The extraction performance processing also situation without weak point.In addition, detection object, contemplated motion are also various, thus characteristic point
Extraction easness it is also different.So as to by that come sets target value, can be tried to achieve by each application or be switched what is matched with situation
Regulating object parameter so that processing cost increases without waste.
Then, the action of information processor 10 that can be realized by above-mentioned structure is illustrated.Figure 14 is to represent
Information processor 10 have rated the processing procedure that is adjusted to adjustment image parameter on the basis of feature point extraction performance
Flow chart.The flow chart starts when camera device 12 have taken evaluation figure.First, the figure image of information processor 10 takes
Obtain the data (S10) that portion 70 obtains the shooting image of the evaluation table figure.
When being set to and apply from the various processing of the shooting of camera device 12 untill information processor 10 obtains data
Equally.That is, on regulating object parameter, also it is given the initial value contemplated in application.Figure image acquiring section 70 will be acquired
Data suitably decoded, extended etc. and being stored to image storage portion 71.Then, feature point extraction portion 72 reads the image
Data, characteristic point (S12) is extracted with the algorithm contemplated in application.
Now, if the number for the characteristic point extracted is not above defined threshold value Th (S14's is no), it is judged to needing
Want the adjustment of regulating object parameter without carrying out performance evaluation, adjustment portion 82 is carried out the adjustment (S22) of parameter.Specifically,
Adjustment portion 82 reads next setting candidate from parameter candidate storage part 84, updates setting in the parameter storage part 64 of camera device 12
It is fixed.Here, threshold value Th setting feature point extraction significant discomforts are closed without the relatively low value with calculation of performance indicators.
If the number for the characteristic point extracted exceedes defined threshold value Th (S14's be), normal solution coordinate obtaining section 74
Normal solution coordinate (S16) is obtained, performance Index Calculation portion 76 calculates the index (S18) of performance evaluation.Specifically, sat according to normal solution
The number of normal solution, the number of non-positive solution calculate F values among the characteristic point that target number, feature point extraction portion 72 are extracted.This
When, performance Index Calculation portion 76 can also by shooting image with normal solution, non-positive solution with different colors, shape to represent
The image for the characteristic point extracted is shown in display device 16.
If the performance indications calculated and the desired value read from desired value storage part 80 are compared by comparing section 78
Result, performance indications are not reaching to desired value (S20's be), then 82 pairs of adjustment image parameters of adjustment portion are adjusted (S22).
Specifically, next setting candidate is read from parameter candidate storage part 84, in the parameter storage part 64 for updating camera device 12
Setting.If carrying out the adjustment of parameter, reflect it and re-fetch the data (S10) of new shooting image that is captured, sending,
Carry out S12~S20 processing.
By it repeatedly, terminate to handle (S20's is no) if the index of the feature point extraction performance in S20 reaches desired value.
Thus in the parameter storage part 64 of camera device 12, setting obtains the treatment conditions of sufficient feature point extraction performance.
Figure 15 represents another example of evaluation figure.Evaluation Figure 30 0, which has, exists the black circle (for example justifying 302) of given size
Arranged on the basis of the space for forming an amount up and down, inside it and gap represents the grey of given size
The pattern of triangle (such as triangle 304).In this case, first, detected by the fitting of circular model from shooting image
The picture of black circle.And as shown in figure right side, try to achieve the centre coordinate (x0, y0) of each circle, (x1, y1) and by they
Coordinate (x1, y0), (x0, the y1) of the intersection point of straight line longitudinally, laterally, are tried to achieve using these coordinates as the three of the given size of center of gravity
Angular apex coordinate is used as normal solution coordinate.
Also, vertex of a triangle is extracted as characteristic point separately through algorithm during application, can be with having used Fig. 5's
Evaluation similarly tries to achieve performance indications with the situation of figure.Like this, evaluation is to exist to try to achieve a part exactly with the pattern of figure
Coordinate algorithm monochromatic figure as benchmark and be expressed as having having for defined position relationship near with the figure
Like represent application when object picture characteristic point shape grey figure combination, shape, configure not by
Limit.
For example Fig. 5 evaluation with the small rectangle in Figure 200 can also be set to triangle, Figure 15 evaluation can also be used
Triangle in Figure 30 0 is set to rectangle.In addition big rectangle in both evaluation figure, circle but lattice-shaped can also be replaced
Straight line etc..Straight line, star, polygon etc. can also be set to instead of small rectangle, triangle.
More than, according to the present embodiment described, using the image that have taken evaluation figure, from the viewpoint of extraction process
Evaluate whether suitably to obtain extracting the image under the stage of characteristic point.In general, from shooting into information processor
The condition of the processing involved by image untill data expansion is optimized arrives good image for people soon.On the other hand, that
The condition of sample is not limited to be adapted under the viewpoint of feature point extraction.So as to from the viewpoint of feature point extraction, to these processing, be made
Equipment gives the leeway of adjustment, so as to more precisely extract characteristic point, and then can also improve the letter using it
Cease the precision of processing.
Adaptability for the system architecture of the extraction of characteristic point quantizes as performance indications are extracted.Now, use
The method different from the feature point extraction algorithm used in application, tries to achieve the coordinate for the characteristic point that should be extracted originally, tries to achieve and add
Enter to extract and extract by mistake and omitted the yardstick of both.Evaluation is set to figure and can try to achieve the spy that should be extracted originally exactly
Levy the corresponding pattern of algorithm of coordinate a little.For example, the small rectangle of the internal representation grey in each rectangle of the pattern of checkerboard.
And accurately detect the corner of the pattern of checkerboard first using general algorithm, asked based on the relative position with it
Obtain the corner coordinate of small rectangle.The characteristic point that algorithm during by it with to apply is extracted is compared, on trying to achieve
State such yardstick.
Even if can have most, condition variation on the influential main cause of feature point extraction processing, it can also pass through
Extraction performance indications are calculated as numerical value and readily compared.In addition device can also be recognized, be determined whether
Adjustment is needed, or is adjusted to improve performance.As a result, can efficiently obtain for using from shooting image
The treatment conditions that the precision for the information processing that the characteristic point extracted is carried out is improved.
More than, the present invention is illustrated based on embodiment.Above-mentioned embodiment is illustrates, and those skilled in the art should manage
Solution can carry out various variations to the combination of these each structural elements, each treatment progress, in addition such variation
It is also at the scope of the present invention.
For example in the present embodiment, in order to suitably obtain characteristic point, to from the shooting of camera device to information processing
The processing carried out untill the expansion of view data in device is adjusted, but the purposes not limited to this of the present invention.For example exist
In the case of being shot with the camera with RGBIR sensors, the result of IR signals is leak into rgb signal, sometimes in RGB
Noise is produced in image.In accordance with the invention it is possible to clearly determine from have taken the image zooming-out of evaluation figure to characteristic point
Among the characteristic point extracted by mistake, so the degree leak into of IR signals can be evaluated by the yield of mistake extraction.
In this case, by using in processing few stage as far as possible to the rgb signal from imaging sensor
Image carry out parameter, so as to caused by the processing with rear class by mistake extract separate.If in addition, using based on IR signals
Image carrys out parameter, then can also evaluate the amount of the noise produced in IR sensors.
Claims (14)
1. a kind of information processor, it is characterised in that possess：
Figure image acquiring section, the data of the shooting image of the evaluation figure of pattern and conduct as defined in representing are obtained from camera device
Image and deploy in memory；
Feature point extraction portion, characteristic point is extracted from the shooting image；
Normal solution coordinate obtaining section, in the method different from the feature point extraction portion, obtains what should be extracted from the shooting image
The position coordinates of characteristic point is used as normal solution coordinate；And
Performance Index Calculation portion, the characteristic point extracted based on the feature point extraction portion and the normal solution coordinate, calculating will be obtained
Suitable for the performance indications after the performance number value of the image of feature point extraction and exported.
2. information processor as claimed in claim 1, it is characterised in that
The extraction algorithm that the evaluation is used comprising the normal solution coordinate obtaining section with the pattern of figure as extracting object monochrome
Figure and the grey that there is defined position relationship with it figure,
The normal solution coordinate obtaining section utilizes the position relationship, obtains the coordinate conduct for the characteristic point that the figure of the grey has
The normal solution coordinate.
3. information processor as claimed in claim 1 or 2, it is characterised in that
The evaluation is with black and white big rectangle of the pattern of figure comprising checkerboard and in the ash shown in the inside of major rectangles
The figure of color.
4. information processor as claimed in claim 1 or 2, it is characterised in that
The pattern of the evaluation figure, which is included, to be configured in multiple black circles in the space of one amount of formation up and down and in each circle
Inside and the figure of grey that represents of the space.
5. the information processor as described in any one of claim 2 to 4, it is characterised in that
The figure of multiple grey in the pattern of the evaluation figure has multiple grey densities.
6. the information processor as described in any one of claim 1 to 5, it is characterised in that be also equipped with：
Adjustment portion, when the performance indications are not reaching to desired value set in advance, to from shooting untill feature point extraction
One of them of setting is adjusted used in the processing for being related to shooting image performed.
7. information processor as claimed in claim 6, it is characterised in that
The desired value make use of the application of the information processing of extracted characteristic point and different according to the rules.
8. the information processor as described in any one of claim 1 to 7, it is characterised in that
The number for the point that the performance Index Calculation portion is extracted by mistake based on the feature point extraction portion and the characteristic point for extracting omission
Number, calculate the performance indications.
9. the information processor as described in any one of claim 1 to 8, it is characterised in that
The performance Index Calculation portion is corresponding with normal solution coordinate among the characteristic point for extracting the feature point extraction portion
Characteristic point makes a distinction in addition with this and represents the image in shooting image and be shown in display device.
10. a kind of evaluation figure, the performance of the image suitable for feature point extraction, its feature are obtained for evaluation information processing unit
It is,
Black and white rectangle comprising the checkerboard for detecting corner from the shooting image and the internal representation in each rectangle
Grey figure, should each rectangle internal representation grey figure have be used for the testing result based on the corner and
Obtain the characteristic point for the results contrast that coordinate and the feature point extraction with being additionally carried out are handled.
11. a kind of evaluation figure, the performance of the image suitable for feature point extraction, its feature are obtained for evaluation information processing unit
It is,
It is configured to comprising what is detected from the shooting image in multiple black circles in the space of one amount of formation up and down and in each circle
Inside and the figure of grey that represents of the space, the figure for the grey that should be represented in the inside of each circle and the space
The results contrast that feature point extraction with coordinate is obtained for the testing result based on the circle and with being additionally carried out is handled
Characteristic point.
12. a kind of evaluation system, it is characterised in that
The camera device of the evaluation figure of defined pattern is represented comprising shooting and obtains the data of the shooting image and evaluates
To the information processor of the performance of the image suitable for feature point extraction,
Described information processing unit possesses：
Figure image acquiring section, obtains the data of the shooting image from camera device and deploys in memory as image；
Feature point extraction portion, characteristic point is extracted from the shooting image；
Normal solution coordinate obtaining section, in the method different from the feature point extraction portion, obtains what should be extracted from the shooting image
The position coordinates of characteristic point is used as normal solution coordinate；
Performance Index Calculation portion, the characteristic point extracted based on the feature point extraction portion and the normal solution coordinate, calculating will be described
Performance indications after performance number value；And
Adjustment portion, when the performance indications are not reaching to desired value set in advance, to the processing institute in the camera device
One of them of the setting used is adjusted,
The camera device possesses：
Parameter storage part, stores the setting being adjusted by the adjustment portion；And
Data forming portion, the Data Concurrent for generating shooting image according to the setting that the parameter storage part is stored delivers to the letter
Cease processing unit.
13. a kind of method of evaluating performance of information processor, it is characterised in that include：
The data of the shooting image of the evaluation figure of pattern as defined in representing are obtained from camera device and as image in storage
The step of deploying in device；
The step of characteristic point being extracted from the shooting image；
The step of with from extracting the characteristic point different method, obtain the position for the characteristic point that should be extracted from the shooting image
The step of coordinate is as normal solution coordinate；And
Based on the characteristic point and the normal solution coordinate extracted in the step of extracting the characteristic point, calculating will obtain being suitable to spy
The step of levying the performance indications after a performance number value for the image extracted and exported.
14. a kind of computer program, it is characterised in that make computer realize following functions：
The data of the shooting image of the evaluation figure of pattern as defined in representing are obtained from camera device and as image in storage
The function of deploying in device；
The function of characteristic point is extracted from the shooting image；
In the method different from the function of extracting the characteristic point, the position for the characteristic point that should be extracted from the shooting image is obtained
Coordinate as normal solution coordinate function；And
Based on the characteristic point and the normal solution coordinate extracted in the function of extracting the characteristic point, calculating will obtain being suitable to spy
Levy the performance indications after a performance number value for the image extracted and the function of being exported.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2016009021A JP2017130794A (en) | 2016-01-20 | 2016-01-20 | Information processing apparatus, evaluation chart, evaluation system, and performance evaluation method |
| JP2016-009021 | 2016-01-20 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107018407A true CN107018407A (en) | 2017-08-04 |
| CN107018407B CN107018407B (en) | 2020-02-07 |
Family
ID=59395751
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710025681.0A Active CN107018407B (en) | 2016-01-20 | 2017-01-13 | Information processing device, evaluation chart, evaluation system, and performance evaluation method |
Country Status (2)
| Country | Link |
|---|---|
| JP (1) | JP2017130794A (en) |
| CN (1) | CN107018407B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108363736A (en) * | 2018-01-19 | 2018-08-03 | 国家测绘地理信息局第三地形测量队 | A kind of storage method, device and the storage system of line entity |
| CN111723641A (en) * | 2019-03-20 | 2020-09-29 | 株式会社理光 | Information processing apparatus, method, system, storage medium, and computer apparatus |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP6933072B2 (en) * | 2017-09-22 | 2021-09-08 | 富士通株式会社 | Camera control method, camera control device and camera control program |
| WO2019093136A1 (en) * | 2017-11-08 | 2019-05-16 | ソニー株式会社 | Image processing device, image processing method, and program |
| US20230035740A1 (en) | 2019-11-21 | 2023-02-02 | Nec Corporation | Parameter determination apparatus, parameter determination method and recording medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101534447A (en) * | 2008-03-13 | 2009-09-16 | 索尼株式会社 | Image processing apparatus and image processing method |
| CN102741882A (en) * | 2010-11-29 | 2012-10-17 | 松下电器产业株式会社 | Image classification device, image classification method, program, recording media, integrated circuit, and model creation device |
| CN102971769A (en) * | 2010-04-30 | 2013-03-13 | 欧姆龙株式会社 | Image transforming device, electronic device, image transforming method, image tranforming program, and recording medium whereupon the program is recorded |
| CN102984429A (en) * | 2011-09-05 | 2013-03-20 | 富士施乐株式会社 | Image processing apparatus, image processing method, and non-transitory computer readable medium |
| CN103813151A (en) * | 2012-11-02 | 2014-05-21 | 索尼公司 | Image processing apparatus and method, image processing system and program |
| WO2015107859A1 (en) * | 2014-01-15 | 2015-07-23 | オムロン株式会社 | Image comparison device, image sensor, processing system, and image comparison method |
-
2016
- 2016-01-20 JP JP2016009021A patent/JP2017130794A/en active Pending
-
2017
- 2017-01-13 CN CN201710025681.0A patent/CN107018407B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101534447A (en) * | 2008-03-13 | 2009-09-16 | 索尼株式会社 | Image processing apparatus and image processing method |
| CN102971769A (en) * | 2010-04-30 | 2013-03-13 | 欧姆龙株式会社 | Image transforming device, electronic device, image transforming method, image tranforming program, and recording medium whereupon the program is recorded |
| CN102741882A (en) * | 2010-11-29 | 2012-10-17 | 松下电器产业株式会社 | Image classification device, image classification method, program, recording media, integrated circuit, and model creation device |
| CN102984429A (en) * | 2011-09-05 | 2013-03-20 | 富士施乐株式会社 | Image processing apparatus, image processing method, and non-transitory computer readable medium |
| CN103813151A (en) * | 2012-11-02 | 2014-05-21 | 索尼公司 | Image processing apparatus and method, image processing system and program |
| WO2015107859A1 (en) * | 2014-01-15 | 2015-07-23 | オムロン株式会社 | Image comparison device, image sensor, processing system, and image comparison method |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108363736A (en) * | 2018-01-19 | 2018-08-03 | 国家测绘地理信息局第三地形测量队 | A kind of storage method, device and the storage system of line entity |
| CN108363736B (en) * | 2018-01-19 | 2022-01-25 | 国家测绘地理信息局第三地形测量队 | Storage method, device and storage system of line entity |
| CN111723641A (en) * | 2019-03-20 | 2020-09-29 | 株式会社理光 | Information processing apparatus, method, system, storage medium, and computer apparatus |
| CN111723641B (en) * | 2019-03-20 | 2024-04-09 | 株式会社理光 | Information processing apparatus, information processing method, information processing system, storage medium, and computer apparatus |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107018407B (en) | 2020-02-07 |
| JP2017130794A (en) | 2017-07-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108537155B (en) | Image processing method, image processing device, electronic equipment and computer readable storage medium | |
| KR102574141B1 (en) | Image display method and device | |
| CN107018407A (en) | Information processor, evaluation figure, evaluation system and method for evaluating performance | |
| JP5395053B2 (en) | Edge mapping using panchromatic pixels | |
| JP5390943B2 (en) | Image processing apparatus and image processing method | |
| US9686537B2 (en) | Noise models for image processing | |
| JP5291084B2 (en) | Edge mapping incorporating panchromatic pixels | |
| CN107563976B (en) | Beauty parameter obtaining method and device, readable storage medium and computer equipment | |
| JP5337049B2 (en) | Reducing position-dependent noise in digital images | |
| JP2004357277A (en) | Digital image processing method | |
| KR20150116833A (en) | Image processor with edge-preserving noise suppression functionality | |
| EP1844441A1 (en) | Determining scene distance in digital camera images | |
| Akyüz | Deep joint deinterlacing and denoising for single shot dual-ISO HDR reconstruction | |
| CN107566684A (en) | Image processing apparatus, image processing method and storage medium | |
| CN109726732A (en) | The system and method for stream for processes data values | |
| CN109729261A (en) | The system and method for stream for processes data values | |
| WO2012008116A1 (en) | Image processing apparatus, image processing method, and program | |
| CN110490196A (en) | Subject detection method and apparatus, electronic equipment, computer readable storage medium | |
| CN109727182A (en) | The system and method for stream for processes data values | |
| WO2008102296A2 (en) | Method for enhancing the depth sensation of an image | |
| CN110996015B (en) | Spatial multiple exposure | |
| CN111080683B (en) | Image processing method, device, storage medium and electronic equipment | |
| JP7030425B2 (en) | Image processing device, image processing method, program | |
| CN109727262A (en) | The system and method for stream for processes data values | |
| CN113824894A (en) | Exposure control method, device, equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |