CN108537739A - A kind of feature based matching degree without reference video enhancing effect evaluating method - Google Patents
A kind of feature based matching degree without reference video enhancing effect evaluating method Download PDFInfo
- Publication number
- CN108537739A CN108537739A CN201810127313.1A CN201810127313A CN108537739A CN 108537739 A CN108537739 A CN 108537739A CN 201810127313 A CN201810127313 A CN 201810127313A CN 108537739 A CN108537739 A CN 108537739A
- Authority
- CN
- China
- Prior art keywords
- frame
- video
- feature
- component
- matching
- 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
- 230000002708 enhancing effect Effects 0.000 title claims abstract description 36
- 238000000034 method Methods 0.000 title claims abstract description 16
- 239000013598 vector Substances 0.000 claims abstract description 24
- 239000000284 extract Substances 0.000 claims abstract description 5
- 230000000694 effects Effects 0.000 claims description 19
- 239000003086 colorant Substances 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 5
- 230000000052 comparative effect Effects 0.000 claims description 2
- 238000000605 extraction Methods 0.000 description 5
- 238000011156 evaluation Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 235000013399 edible fruits Nutrition 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000013139 quantization Methods 0.000 description 2
- 238000004040 coloring Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 230000003389 potentiating effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/90—Dynamic range modification of images or parts thereof
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
- G06V10/757—Matching configurations of points or features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Multimedia (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Databases & Information Systems (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to a kind of feature based matching degrees without reference video enhancing effect evaluating method, includes the following steps:Quality enhancing is carried out to the original video V containing N frames using a certain video enhancement algorithm E, has been enhanced video Ve;Being obtained successively by frame sequential has enhanced the primary color component image of video frame, extracts the feature vector of component image respectively using invariant features operator;Characteristic matching is carried out to the feature vector of same primary color component in front and back frame successively, calculates separately out the two successful numbers of width component image Feature Points Matching;Judge whether present frame is to have enhanced video VeLast frame, if it is not, then return above-mentioned steps;Otherwise, into next step;The characteristic matching points of accumulative all front and back frame primary color components, every frame component average value that all characteristic matchings are counted is as the characteristic matching degree of the video enhancement algorithm.The present invention provides a kind of objective evaluating criterion to improve without reference video enhancing algorithm.
Description
Technical field
The present invention relates to technical field of image processing, increase without reference video more particularly to a kind of feature based matching degree
Potent fruit evaluating method.
Background technology
In machine vision applications, the feature extraction of image is the basic problem of image procossing, and accurate efficient feature is calculated
Son provides solid underlying basis for the solution of relevant issues.Feature operator based on scale space is by original image
Change of scale is carried out, the expression sequence under Image Multiscale is obtained, these sequences are carried out with the extraction of scale space main outline, and
Using the main outline as a kind of invariant features.Such invariant features operator is not necessarily to the priori of image, can realize side
Feature extraction on edge, Corner Detection and different resolution.In recent years, with Scale invariant features transform (Scale Invariant
Feature Transform, SIFT), accelerate robust features (Speeded Up Robust Features, SURF) be representative
Invariant features operator be gradually widely used in various machine vision and area of pattern recognition, utilize invariant features operator
Extraction is characterized in the local feature of image, maintains the invariance to rotation, scaling, brightness change.
Machine vision applications may face low-quality original video scene, and can not obtain other scene informations, this is just
Need to introduce the video source modeling mechanism of no reference.No reference video enhancing algorithm is to add one to original video by certain means
A little information or transformation data selectively protrude interested feature in video or inhibit certain unwanted spies in video
Sign expands the difference between different objects feature in video, to meet certain special demands.Currently, enhancing without reference video
The effect evaluating mechanism of algorithm is perfect not enough, it is difficult to a kind of objective evaluating criterion of suitable machine vision applications is provided, with card
Bright a certain video enhancement algorithm has better robustness and validity, the conjunction which has limited related application to video enhancement algorithm
Reason selection.
It was found by the inventors of the present invention that since the picture material of video consecutive frame usually has very strong relevance, it is adjacent
Largely variation is smaller for the content for including between frame, and the matching degree of consecutive frame image feature can preferably embody video source modeling
The effect of algorithm.Variation of the enhanced video image relative to original image, mainly image local feature point increase.Profit
With invariant features operator, same or analogous target in two images can be quickly found, this is commented for video source modeling effect
Survey provides a kind of new thinking.How without reference to video, video source modeling is measured according to objective criteria
The technical issues of effect is urgent need to resolve.
Invention content
Technical problem to be solved by the invention is to provide a kind of feature based matching degrees without reference video enhancing effect
Evaluating method can evaluate application effect of no reference video enhancing algorithm in machine vision.
The technical solution adopted by the present invention to solve the technical problems is:There is provided a kind of feature based matching degree without reference
Video source modeling effect evaluating method, includes the following steps:
(1) quality enhancing is carried out to the original video V containing N frames using a certain video enhancement algorithm E, has been enhanced and has been regarded
Frequency Ve;
(2) being obtained successively by frame sequential has enhanced the primary color component image of video frame, is distinguished using invariant features operator
Extract the feature vector of component image;
(3) characteristic matching is carried out to the feature vector of same primary color component in front and back frame successively, calculates separately out two width point
Measure the successful number of Image Feature Point Matching;
(4) judge whether present frame is to have enhanced video VeLast frame, if it is not, then going to step (2);It is no
Then, it enters step (5);
(5) add up the characteristic matching points of all front and back frame primary color components, every frame component that all characteristic matchings are counted
Characteristic matching degree of the average value as the video enhancement algorithm.
The step (2) is specially:Reading has enhanced video Ve, demarcate frame number;Being obtained successively by frame number has enhanced
Video VeIn the i-th frame R, G, B three primary colours component image, three kinds of component images are carried out with the positioning and description of main outline respectively,
The feature for being extracted i-th three kinds of component images of frame respectively using invariant features operator, is obtained respectively by scale space extremum extracting
The position of component image initial characteristics point and dimensional information, thus to obtain the feature vector of each component image of the i-th frame
The invariant features operator is SIFT operators or SURF operators.
The step (3) is specially:Enable the (i-1)-th frame be previous frame image, the i-th frame be a later frame image, then the (i-1)-th frame and
I-th frame forms jth to front and back frame, and characteristic matching is carried out using the feature vector of same primary color component in the (i-1)-th frame and the i-th frame,
I.e. in gainedWithWithWithIn matched one by one, calculate separately two width component image characteristic points
The number of successful match obtains jth and counts the characteristic matching of each primary color component of front and back frame after removing erroneous matchingIf successful Feature Points Matching, spy of the jth to front and back frame primary color component is not present in a certain primary color component
Sign matching pointsOrOr
The matching double points that mistake is eliminated using RANSAC algorithm, according to the nearest neighbor distance of comparative feature vector
Characteristic point is matched with the mode of secondary nearest neighbor distance, nearest neighbor distance and the distance rates of time nearest neighbor distance are less than threshold epsilon
When, then it is assumed that it is correct matching double points.
The step (5) is specially:The characteristic matching points of accumulative all front and back frames of three primary colours component, finally calculate all
Every frame component average value of characteristic matching points, finding out video enhancement algorithm E is enhancing video VeOn characteristic matching degreeCharacteristic matching degree D is the foundation for evaluating and testing video enhancement algorithm, and the higher expression of D values should
The effect of video enhancement algorithm is more preferable.
Advantageous effect
Due to the adoption of the above technical solution, compared with prior art, the present invention having the following advantages that and actively imitating
Fruit:The present invention obtains the characteristic matching points for having enhanced front and back frame component image in video using invariant features operator, in turn
The good and bad degree of various no reference video enhancing algorithms is evaluated and tested by characteristic matching degree.Carried criterion is regarded without necessarily referring to original
Frequently, the characteristic matching degree index for being capable of providing fine quantization carries out the nothing of video source modeling effect and participates in evaluation and electing survey, is kept away in test process
Parameter adjustment is exempted from, has been not necessarily to manual intervention, suitable for the application scenarios of no reference video enhancing, especially various machine vision are answered
With.
Description of the drawings
Fig. 1 is the flow chart that criterion is evaluated and tested without reference video enhancing effect of feature based matching degree.
Fig. 2 is the various enhancing algorithm effect exemplary plots using SIFT feature matching degree.
Fig. 3 is the various enhancing algorithm effect exemplary plots using SURF characteristic matching degree.
Specific implementation mode
Present invention will be further explained below with reference to specific examples.It should be understood that these embodiments are merely to illustrate the present invention
Rather than it limits the scope of the invention.In addition, it should also be understood that, after reading the content taught by the present invention, people in the art
Member can make various changes or modifications the present invention, and such equivalent forms equally fall within the application the appended claims and limited
Range.
Embodiments of the present invention be related to a kind of feature based matching degree without reference video enhancing effect evaluating method, packet
Include following steps:Quality enhancing is carried out to the original video V containing N frames using a certain video enhancement algorithm E, has been enhanced and has been regarded
Frequency Ve;Being obtained successively by frame sequential has enhanced the primary color component image of video frame, is extracted and is divided respectively using invariant features operator
The feature vector of spirogram picture;Characteristic matching is carried out to the feature vector of same primary color component in front and back frame successively, is calculated separately out
The two successful numbers of width component image Feature Points Matching;Judge whether present frame is to have enhanced video VeLast frame, if
It is not then to return to above-mentioned steps;Otherwise, into next step;The characteristic matching points of accumulative all front and back frame primary color components, will
Characteristic matching degree of the every frame component average value of all characteristic matching points as the video enhancement algorithm.
It is further illustrated the present invention below by specific embodiment.
In the present embodiment, six kinds without reference video enhancing algorithm be respectively DeepDetail algorithms, Derain algorithms,
MRF algorithms, NLD algorithms, DCP algorithms, SP algorithms, they are respectively to same original video Hazyvideo_riverside.avi
Quality enhancing is carried out, the above method is then respectively adopted and calculates the characteristic matching degree for having enhanced video, there is higher characteristic matching
The enhancing algorithm of degree will be judged as having better enhancing effect.For convenience, setting video enhancing algorithm E is represented above-mentioned
Any one of six kinds of video enhancement algorithms, Fig. 1 give the video source modeling effect evaluation and test criterion for implementing feature based matching degree
Flow chart, carried criterion pass through step in detail below realize:
Step 1:The video HazyVideo_riverside.avi containing 300 frames is carried out using video enhancement algorithm E
Quality enhances, and after all picture frames enhance in HazyVideo_riverside.avi, has been enhanced video Ve。
Step 2:Reading has enhanced video Ve, demarcate frame number.Video V is obtained firsteIn the 1st frame R, G, B three primary colours
Component image carries out the R component image, G component images, B component image of the frame positioning and description of main outline respectively, utilizes
Invariant features operator extracts the SIFT feature or SURF features of three kinds of component images of the 1st frame respectively, passes through scale space extreme value
Detection obtains position and the dimensional information of each primary color component initial characteristics point, and obtains the feature vector of each component image respectively
Step 3:Obtain video VeIn the i-th frame (i >=2) R, G, B three primary colours component image, calculated using invariant features
Son extracts the SIFT feature or SURF features of three kinds of component images of the i-th frame respectively, and the frame is obtained by scale space extremum extracting
The position of each primary color component initial characteristics point and dimensional information, thus obtain the feature vector of each component image respectively
Step 4:(i-1)-th frame image and the i-th frame image form jth to frame before and after (j=i-1), successively to acquisition before and after
Frame primary color component is handled, and characteristic matching is carried out using the feature vector of same primary color component in front and back frame.Carried criterion is adopted
With the SIFT feature vector or SURF feature vectors of each component image of invariant features operator extraction.Feature vector is between any two
Euclidean distance passes through formulaObtain, wherein p (x, y) indicate feature vector x and feature vector y it
Between Euclidean distance, n indicate feature vector n-th dimension.Carried criterion is in gainedWithWithWith
In carry out characteristic matching one by one, the matching double points typically resulted in can all have the matching of mistake, therefore be gone using RANSAC algorithms
Except the matching double points of mistake, calculate the nearest neighbor distance and time nearest neighbor distance between feature vector, if nearest neighbor distance with it is time close
The ratio of neighborhood distance is then considered correct matching double points less than threshold epsilon=0.5, to extract correct match point with a high credibility
It is right.The same successful number of primary color component Feature Points Matching of the front and back frame of carried criterion statistics, respectively obtains jth to each base of front and back frame
The characteristic matching of colouring component is countedPer the characteristic point of a pair of successful match, indicate the two characteristic points in reality
Belong to same content in the scene of border.
Step 5:If not reaching has enhanced video VeLast frame, then continue to obtain successively by frame number latter
Frame, i.e. i=i+1, go to step 3;Conversely, into last step six.
Step 6:The characteristic matching points of accumulative all front and back frames, using the size of the sum of characteristic matching points as effect
The generation foundation of scoring calculates every frame component average value of all characteristic matching points, using formulaVideo enhancement algorithm E is found out in video VeOn SIFT or SURF characteristic matchings
Degree.Fig. 2 gives the various enhancing algorithm effect examples using SIFT feature matching degree, the experimental results showed that NLD video source modelings
Algorithm has the effect of best.Fig. 3 gives the various enhancing algorithm effect examples using SURF characteristic matching degree, experimental result
It is best to show that DCP video enhancement algorithms have the effect of.In this example, NLD algorithms and DCP algorithms obtain higher D values,
Therefore there is relatively better enhancing effect.
It is not difficult to find that the present invention provides a kind of objective evaluating criterion to improve without reference video enhancing algorithm.The present invention
Being obtained using invariant features operator has enhanced the characteristic matching points of front and back frame component image in video, and then passes through feature
Various no reference videos are evaluated and tested with degree enhances the good and bad degree of algorithm.Carried criterion is without necessarily referring to original video, Neng Gouti
It carries out the nothing of video source modeling effect for the characteristic matching degree index of fine quantization and participates in evaluation and electing survey, parameter tune is avoided in test process
It is whole, it is not necessarily to manual intervention, suitable for the application scenarios of no reference video enhancing, especially various machine vision applications.
Claims (6)
1. a kind of feature based matching degree without reference video enhancing effect evaluating method, which is characterized in that include the following steps:
(1) quality enhancing is carried out to the original video V containing N frames using a certain video enhancement algorithm E, has been enhanced video Ve;
(2) being obtained successively by frame sequential has enhanced the primary color component image of video frame, is extracted respectively using invariant features operator
The feature vector of component image;
(3) characteristic matching is carried out to the feature vector of same primary color component in front and back frame successively, calculates separately out two width component maps
As the successful number of Feature Points Matching;
(4) judge whether present frame is to have enhanced video VeLast frame, if it is not, then going to step (2);Otherwise, enter
Step (5);
(5) the characteristic matching points for adding up all front and back frame primary color components, every frame component that all characteristic matchings are counted is averaged
It is worth the characteristic matching degree as the video enhancement algorithm.
2. feature based matching degree according to claim 1 without reference video enhancing effect evaluating method, feature exists
In the step (2) is specially:Reading has enhanced video Ve, demarcate frame number;Being obtained successively by frame number has enhanced video Ve
In the i-th frame R, G, B three primary colours component image, three kinds of component images are carried out with the positioning and description of main outline respectively, using not
Characteristics of variables operator extracts the feature of i-th three kinds of component images of frame respectively, and each component map is obtained by scale space extremum extracting
Position as initial characteristics point and dimensional information, thus to obtain the feature vector of each component image of the i-th frame
3. feature based matching degree according to claim 2 without reference video enhancing effect evaluating method, feature exists
In the invariant features operator is SIFT operators or SURF operators.
4. feature based matching degree according to claim 1 without reference video enhancing effect evaluating method, feature exists
In the step (3) is specially:It is previous frame image to enable the (i-1)-th frame, and the i-th frame is a later frame image, then the (i-1)-th frame and the i-th frame
Jth is formed to front and back frame, carries out characteristic matching using the feature vector of same primary color component in the (i-1)-th frame and the i-th frame, i.e., in institute
WithWithWithIn matched one by one, calculate separately two width component image Feature Points Matchings
Successful number obtains jth and counts the characteristic matching of each primary color component of front and back frame after removing erroneous matchingIf successful Feature Points Matching, spy of the jth to front and back frame primary color component is not present in a certain primary color component
Sign matching pointsOrOr
5. feature based matching degree according to claim 4 without reference video enhancing effect evaluating method, feature exists
In, the matching double points of mistake are eliminated using RANSAC algorithm, according to the nearest neighbor distance of comparative feature vector with it is secondary
The mode of nearest neighbor distance matches characteristic point, when nearest neighbor distance and the distance rates of time nearest neighbor distance are less than threshold epsilon, then
It is considered correct matching double points.
6. feature based matching degree according to claim 1 without reference video enhancing effect evaluating method, feature exists
In the step (5) is specially:The characteristic matching points of accumulative all front and back frames of three primary colours component, finally calculate all features
Every frame component average value of points is matched, finding out video enhancement algorithm E is enhancing video VeOn characteristic matching degreeCharacteristic matching degree D is the foundation for evaluating and testing video enhancement algorithm, and the higher expression of D values should
The effect of video enhancement algorithm is more preferable.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810127313.1A CN108537739B (en) | 2018-02-08 | 2018-02-08 | Non-reference video enhancement effect evaluation method based on feature matching degree |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810127313.1A CN108537739B (en) | 2018-02-08 | 2018-02-08 | Non-reference video enhancement effect evaluation method based on feature matching degree |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108537739A true CN108537739A (en) | 2018-09-14 |
CN108537739B CN108537739B (en) | 2021-11-09 |
Family
ID=63485436
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810127313.1A Active CN108537739B (en) | 2018-02-08 | 2018-02-08 | Non-reference video enhancement effect evaluation method based on feature matching degree |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108537739B (en) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102868907A (en) * | 2012-09-29 | 2013-01-09 | 西北工业大学 | Objective evaluation method for quality of segmental reference video |
CN103674968A (en) * | 2013-12-20 | 2014-03-26 | 纪钢 | Method and device for evaluating machine vision original-value detection of exterior corrosion appearance characteristics of material |
CN103996192A (en) * | 2014-05-12 | 2014-08-20 | 同济大学 | Non-reference image quality evaluation method based on high-quality natural image statistical magnitude model |
US20150213585A1 (en) * | 2014-01-30 | 2015-07-30 | Canon Kabushiki Kaisha | Image processing apparatus, image processing method, and storage medium |
CN105006001A (en) * | 2015-08-19 | 2015-10-28 | 常州工学院 | Quality estimation method of parametric image based on nonlinear structural similarity deviation |
-
2018
- 2018-02-08 CN CN201810127313.1A patent/CN108537739B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102868907A (en) * | 2012-09-29 | 2013-01-09 | 西北工业大学 | Objective evaluation method for quality of segmental reference video |
CN103674968A (en) * | 2013-12-20 | 2014-03-26 | 纪钢 | Method and device for evaluating machine vision original-value detection of exterior corrosion appearance characteristics of material |
US20150213585A1 (en) * | 2014-01-30 | 2015-07-30 | Canon Kabushiki Kaisha | Image processing apparatus, image processing method, and storage medium |
CN103996192A (en) * | 2014-05-12 | 2014-08-20 | 同济大学 | Non-reference image quality evaluation method based on high-quality natural image statistical magnitude model |
CN105006001A (en) * | 2015-08-19 | 2015-10-28 | 常州工学院 | Quality estimation method of parametric image based on nonlinear structural similarity deviation |
Also Published As
Publication number | Publication date |
---|---|
CN108537739B (en) | 2021-11-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Wen et al. | COVERAGE—A novel database for copy-move forgery detection | |
Nafchi et al. | Efficient no-reference quality assessment and classification model for contrast distorted images | |
CN104850850B (en) | A kind of binocular stereo vision image characteristic extracting method of combination shape and color | |
Chen et al. | Image splicing detection via camera response function analysis | |
CN104462381A (en) | Trademark image retrieval method | |
CN109472770B (en) | Method for quickly matching image characteristic points in printed circuit board detection | |
US20080292200A1 (en) | Visual enhancement for reduction of visual noise in a text field | |
Shahroudnejad et al. | Copy-move forgery detection in digital images using affine-SIFT | |
CN115244542A (en) | Method and device for verifying authenticity of product | |
CN110136125B (en) | Image copying and moving counterfeiting detection method based on hierarchical feature point matching | |
CN104657980A (en) | Improved multi-channel image partitioning algorithm based on Meanshift | |
CN111126412A (en) | Image key point detection method based on characteristic pyramid network | |
Khalid et al. | Bhattacharyya Coefficient in Correlation of Gray-Scale Objects. | |
CN103080979A (en) | System and method for synthesizing portrait sketch from photo | |
CN104123554A (en) | SIFT image characteristic extraction method based on MMTD | |
CN106682679A (en) | Significance detection method based on level-set super pixel and Bayesian framework | |
US20110085026A1 (en) | Detection method and detection system of moving object | |
CN109635679B (en) | Real-time target paper positioning and loop line identification method | |
CN114913463A (en) | Image identification method and device, electronic equipment and storage medium | |
Julliand et al. | Automated image splicing detection from noise estimation in raw images | |
CN108537739A (en) | A kind of feature based matching degree without reference video enhancing effect evaluating method | |
CN106056575A (en) | Image matching method based on object similarity recommended algorithm | |
CN112184533B (en) | Watermark synchronization method based on SIFT feature point matching | |
CN111402281B (en) | Book edge detection method and device | |
CN108876849B (en) | Deep learning target identification and positioning method based on auxiliary identification |
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 |