CN110263205A - A kind of search method for ginseng image - Google Patents
A kind of search method for ginseng image Download PDFInfo
- Publication number
- CN110263205A CN110263205A CN201910491812.3A CN201910491812A CN110263205A CN 110263205 A CN110263205 A CN 110263205A CN 201910491812 A CN201910491812 A CN 201910491812A CN 110263205 A CN110263205 A CN 110263205A
- Authority
- CN
- China
- Prior art keywords
- image
- ginseng
- binary map
- obtains
- original
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
- G06F16/5838—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content using colour
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Library & Information Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of search methods for ginseng image.Original ginseng image is the image of RGB color, and the image handled is the ginseng image for removing most of background；Hsv color space is gone to from RGB color again, processing obtains third binary map and carries out obtaining ginseng image with operation as mask and original ginseng image；Ginseng in ginseng image must be removed using morphological operation, carry out rotation correction direction, then ginseng image is cut so that ginseng is full of whole image, carry out uniform piecemeal, calculate the ratio that ginseng target in each image block accounts for whole image, the weight of image block is obtained, processing obtains Hash coding；Then it is encoded according to Hash and calculates similarity distance, retrieval process is carried out according to similarity distance.The present invention can quickly and efficiently retrieve ginseng image, have preferable retrieval rate.
Description
Technical field
The present invention relates to the methods of medicinal material ginseng image retrieval, are specifically related to a kind of Hash coding based on image block
Image search method.
Background technique
Ginseng is as a kind of rare medicinal material, and very popular on the market, different ginsengs have different grades, grade
Between drug effect and price all differ bigger, so for ginseng carry out grade it is correct identify be particularly important.People
The level ratings of ginseng are to need the professional of food office, according to the evaluation criterion of the kind of ginseng and different cultivars to its into
Row grading.Grading needs multidigit professional to carry out comprehensive comparation and assessment.It, may for a ginseng for having commented grade
The phenomenon that in the presence of multiple grading.If buyer distrusts the ginseng grade that need to be bought, so may require that professional rating organization again
It is secondary to grade.
There are the repeatability of work for current repeatability grading work, while may lead there is also repeatedly grading inconsistent
The professional problem caused.
Summary of the invention
In order to solve the problems, such as to carry out repeatability grading to ginseng, the present invention provides a kind of pair of ginseng images to retrieve
Method.This method can avoid repeatedly grading task, reduce workload, while also can avoid repeatedly grade it is inconsistent caused by
Professional problem.
Institute of the invention is included the following steps: using technical solution
1) original ginseng image is the image of RGB color, goes to hsv color from RGB color to ginseng image
Space carries out the filtering of color to image, image progress binaryzation is then obtained the first binary map again, the binary map that will be obtained
It carries out and operates with original ginseng image, obtained image is the ginseng image for removing most of background；
The original ginseng image is to be acquired to obtain by camera, has complete single ginseng in image, and background is single
One color background.
2) the ginseng image of the most of background of removal obtained in step 1) is gone into HSV from RGB color again
Color space is further filtered the pixel lower than hsv color threshold value using hsv color threshold value, then carries out two-value
Change obtains the second binary map, and largest connected domain, other companies of removal other than largest connected domain are searched in the second binary map
Logical domain, obtains third binary map, carries out and operates, obtained image using third binary map as mask and original ginseng image
For the ginseng image after final background removal；
3) ginseng in the ginseng image for obtaining step 2) must be removed using morphological operation；
4) the ginseng image for the tiny root of removal that step 3) obtains is subjected to rotation correction direction, then to ginseng image
It is cut so that ginseng is full of whole image；
5) the ginseng image in step 4) is subjected to uniform piecemeal, calculates ginseng target in each image block and accounts for entirely
The ratio of image, obtains the weight of image block, and processing obtains Hash coding；Then it is encoded according to Hash and calculates similarity distance,
Retrieval process is carried out according to similarity distance.
The color filtering is under the color space of HSV, and the threshold value of setting H, S, V carry out color filtering.
Specific step is as follows for the step 3):
3.1) the ginseng image after background removal that step 2) obtains is subjected to binarization operation and obtains the 4th binary map, then
It carries out top cap operation and obtains top cap operating result figure, obtain the tiny root part of ginseng in top cap operating result figure；
3.2) top cap operating result figure and the 4th binary map are subjected to xor operation, obtain exclusive or result figure, exclusive or result
It is the ginseng binary map for eliminating tiny root in figure；
3.3) the largest connected domain in exclusive or result figure is searched, is to eliminate the tiny root of ginseng in the maximum UNICOM domain
Largest connected domain and original ginseng image are carried out the ginseng that the final tiny root of removal is obtained with operation by bianry image region
Image.
Specific step is as follows for the step 4):
4.1) gray processing will be carried out through step 3) treated ginseng image first and obtain grayscale image, then grayscale image will be carried out
Binarization operation obtains the 5th binary map；
4.2) largest connected region is found in the 5th binary map, one will be established using the largest connected region as profile
The minimum rectangular area in the largest connected region is surrounded, and minimum rectangular area is drawn to form minimum in original ginseng image
Rectangle surrounds frame, and minimum rectangle is calculated and surrounds the center point coordinate of frame and the long side and image water of minimum rectangle encirclement frame
Angle theta square between rotates the long side that original ginseng image makes minimum rectangle surround frame around center point coordinate and is parallel to
Image level direction；
4.3) frame is surrounded using minimum rectangle to cut original ginseng image, retain minimum rectangle and surround within the scope of frame
Image-region, by minimum rectangle surround frame other than original ginseng image image-region delete, so that ginseng target is full of
Whole image.
Specific step is as follows for the step 5):
5.1) size of the ginseng image obtained to step 4) zooms in and out unified 400 × 200 size, then will figure
As being uniformly divided into muti-piece image block, specific implementation divides the image into 8 pieces etc. big of ginseng image block, the size of each piece of image block
It is 100 × 100 size；
5.2) ginseng image weights in each image block are calculated:
Processing obtains the area of ginseng part in area S and each image block occupied by ginseng part in whole image
di, i=1,2 ..., 8, i indicate image block serial number, and then are calculated what ginseng Partial Feature in each piece of image block was extracted
Weighted value Wi=di/ S, the ginseng image weights as each image block；
The calculating of area carries out binarization operation especially by ginseng image, by searching for the method for connected domain, statistics
The sized images region area of connected domain.
5.3) each image block is narrowed down into uniform sizes size, the detail section of picture can be removed, retain entire piecemeal
The image structure information of image, total pixel of such image block are 256 pixels；Image block after diminution is carried out at binaryzation
Reason, is then assigned a value of 1 for the pixel of ginseng main part, and the pixel in other regions is assigned a value of 0, and building is obtained by 0,1 number
It is worth the binary map of composition, establishes the sequence of pixel values being made of 0,1 numerical value and encoded as the Hash of the image block, Hash coding
The sequence constituted by 256 0,1 forms, and the Hash coding of each image block is arranged according to the combination order of image block；
5.4) each image in the image library for inputting ginseng image and inquiry to be retrieved, is all made of aforesaid way
Obtain image block and its Hash coding；Ginseng image to be retrieved is inputted, each of which image block is calculated according to Hamming distance and looks into
Similarity distance in the image library of inquiry between the image block of each image, by the similarity distance and corresponding diagram of all image blocks
As the product of the ginseng image weights of block is added, as final similarity distance, finally according to being ranked up from small to large, distance
It is the smallest as the most like ginseng image retrieved.
The beneficial effects of the present invention are:
1) by carrying out interested extracted region to ginseng image in the present invention, so that the retrieval of ginseng image is not carried on the back
The influence of scape improves the accuracy of retrieval.
2) present invention retrieves the retrieval of ginseng image using the method for piecemeal Hash coding, and the speed of retrieval compares
Fastly, there is preferable real-time.
3) the method for the present invention has wide range of applications, and method simplicity is easy to apply, and can be applied to other searching fields, tool
There is biggish application potential.
For synthesis, the present invention can quickly and efficiently be retrieved ginseng image, have preferable retrieval accurate
Rate.
Detailed description of the invention
Fig. 1 is flow chart of the invention.
Fig. 2 is the flow chart that the removal of ginseng image background is carried out using HSV.
Fig. 3 is the flow chart for carrying out root removal to ginseng using morphological operation.
Fig. 4 is that the present invention encodes the flow chart retrieved using piecemeal Hash.
Fig. 5 is the ginseng image of retrieval.Wherein: a is ginseng original image；B is to be transformed into ginseng original image from RGB color
The result figure in hsv color space.
Fig. 6 is the procedure chart that background removal is carried out to ginseng image.Wherein: a is that the ginseng after HSV threshold filtering shows
It is intended to；B is by the schematic diagram of the ginseng binaryzation after color filtering；C is showing for original image and largest connected domain binary map and operation
It is intended to；D is the final ginseng effect picture of HSV background removal.
Fig. 7 is the procedure chart that root removal is carried out to ginseng image.Wherein: a is ginseng original image；B is two after binaryzation
Value figure；C is that the result figure after top cap operation is carried out to binary map；D is the result figure after binary map and top cap figure exclusive or；E is to look into
Look for the result figure of maximum connected domain；F is the result figure removed after ginseng palpus.
Fig. 8 is the procedure chart that rotation and dimension correction are carried out to ginseng image.Wherein: a is after removing background and root
Result figure；B is the result figure that binaryzation is carried out to ginseng；C is to calculate the minimum rectangle that ginseng region is formed in figure to surround frame；d
To carry out the result figure after rotation correction to ginseng image；E is the result figure after cutting to ginseng region；F is to ginseng
Scale zooms in and out the result figure after the size of setting.
Fig. 9 is the result figure that uniform piecemeal is carried out to ginseng image.
Figure 10 is to the result figure after ginseng block image binaryzation.
Specific embodiment
The present invention is further illustrated in the following with reference to the drawings and specific embodiments.
As shown in Figure 1, implementation process of the invention is as follows:
Step 1: original ginseng image is the image of RGB color, goes to HSV from RGB color to ginseng image
Color space carries out the filtering of color to image, image progress binaryzation is then obtained the first binary map again, two will obtained
Value figure is carried out and is operated with original ginseng image, and obtained image is the ginseng image for removing most of background, and flow chart is such as
Shown in Fig. 2.
Step 2: the ginseng image of the most of background of removal obtained in step 1 is turned from RGB color again
To hsv color space, it further is filtered the pixel lower than hsv color threshold value using hsv color threshold value, is then carried out
Binaryzation obtains the second binary map, and largest connected domain is searched in the second binary map, removes its other than largest connected domain
His connected domain, obtains third binary map, carries out and operates using third binary map as mask and original ginseng image, obtains
Image is the ginseng image after final background removal.
Step 3: the root of the ginseng in step 2 is gone using morphological operation, process flow diagram is as shown in Figure 3；Tool
Body process are as follows:
3.1) the ginseng image after background removal that step 2 obtains is subjected to binarization operation and obtains the 4th binary map, then
It carries out top cap operation and obtains top cap operating result figure, obtain the tiny root part of ginseng in top cap operating result figure；
3.2) top cap operating result figure and the 4th binary map are subjected to xor operation, obtain exclusive or result figure, exclusive or result
It is the ginseng binary map for eliminating tiny root in figure；
3.3) the largest connected domain in exclusive or result figure is searched, is to eliminate the tiny root of ginseng in the maximum UNICOM domain
Largest connected domain and original ginseng image are carried out the ginseng that the final tiny root of removal is obtained with operation by bianry image region
Image.
Step 4: the removal ginseng image in step 3 is subjected to rotation correction direction, then ginseng image is cut out
It cuts so that ginseng is full of whole image；Detailed process are as follows:
4.1) gray processing will be carried out through step 3 treated ginseng image first and obtain grayscale image, then grayscale image will be carried out
Binarization operation obtains the 5th binary map；
4.2) largest connected region is found in the 5th binary map, one will be established using the largest connected region as profile
The minimum rectangular area in the largest connected region is surrounded, and minimum rectangular area is drawn to form minimum in original ginseng image
Rectangle surrounds frame, and minimum rectangle is calculated and surrounds the center point coordinate of frame and the long side and image water of minimum rectangle encirclement frame
Angle theta square between rotates the long side that original ginseng image makes minimum rectangle surround frame around center point coordinate and is parallel to
Image level direction；
4.3) frame is surrounded using minimum rectangle to cut original ginseng image, retain minimum rectangle and surround within the scope of frame
Image-region, by minimum rectangle surround frame other than original ginseng image image-region delete, so that ginseng target is full of
Whole image.
Step 5: the ginseng image in step 4) is subjected to uniform piecemeal, calculates ginseng image in each piecemeal
The ratio for accounting for entire ginseng image obtains the weight of piecemeal ginseng image, then piecemeal is carried out binary conversion treatment, will be in binary map
Value encoded as the Hash of piecemeal ginseng, flow chart is as shown in Figure 4；Detailed process are as follows:
5.1) size of the ginseng image obtained to step 4) zooms in and out unified 400 × 200 size, then will figure
As being uniformly divided into muti-piece image block, specific implementation divides the image into 8 pieces etc. big of ginseng image block, the size of each piece of image block
It is 100 × 100 size；
5.2) ginseng image weights in each image block are calculated:
Processing obtains the area of ginseng part in area S and each image block occupied by ginseng part in whole image
di, i=1,2 ..., 8, i indicate image block serial number, and then are calculated what ginseng Partial Feature in each piece of image block was extracted
Weighted value Wi=di/ S, the ginseng image weights as each image block；
The calculating of area carries out binarization operation especially by ginseng image, by searching for the method for connected domain, statistics
The sized images region area of connected domain.
5.3) each image block is narrowed down into uniform sizes size, the detail section of picture can be removed, retain entire piecemeal
The image structure information of image, total pixel of such image block are 256 pixels；Image block after diminution is carried out at binaryzation
Reason, is then assigned a value of 1 for the pixel of ginseng main part, and the pixel in other regions is assigned a value of 0, and building is obtained by 0,1 number
It is worth the binary map of composition, establishes the sequence of pixel values being made of 0,1 numerical value and encoded as the Hash of the image block, Hash coding
The sequence constituted by 256 0,1 forms, and the Hash coding of each image block is arranged according to the combination order of image block；
5.4) each image in the image library for inputting ginseng image and inquiry to be retrieved, is all made of aforesaid way
Obtain image block and its Hash coding；Ginseng image to be retrieved is inputted, each of which image block is calculated according to Hamming distance and looks into
Similarity distance in the image library of inquiry between the image block of each image, by the similarity distance and corresponding diagram of all image blocks
As the product of the ginseng image weights of block is added, as final similarity distance, finally according to being ranked up from small to large, distance
It is the smallest as the most like ginseng image retrieved.
Specific embodiment:
The present invention, which is used, to be schemed such as a in Fig. 5 shown in as original input picture of the invention, by the original image of input from
Rgb space is transformed into HSV space, obtains the HSV space image as shown in b figure in Fig. 5.After obtaining the image of HSV, setting H, S,
The threshold value of V filters the background color of HSV space image, the image for removing most of background is obtained, as shown in a in Fig. 6；Then
The image that most of background will be removed carries out binarization operation, then searches largest connected domain, obtains only first binary map, by the
One binary map and original image carry out obtaining figure shown in c in Fig. 6 with operation.Then c figure in Fig. 6 is subjected to HSV face again
The ginseng image for eliminating background in Fig. 6 shown in d is obtained by filtration in chromatic threshold value.It is illustrated in figure 7 and root removal is carried out to ginseng
Processing then operate to obtain the top cap operation of the root part of ginseng using top cap by carrying out binary conversion treatment to ginseng
B figure in Fig. 7 is carried out xor operation with top cap operating result figure and obtains d figure in Fig. 7, then by searching for maximum by result figure
The method of connected domain has obtained the result figure after removal ginseng palpus, as shown in f in Fig. 7.Fig. 8 is the knot to removal ginseng palpus
Fruit figure carries out the procedure chart of rotation with dimension correction, and the result figure for removing ginseng palpus is carried out binaryzation, is then searched maximum
Connected domain obtains the maximum region in binary map, will be using the largest connected region as profile, and establishing one, to surround this largest connected
The minimum rectangular area in region, and draw minimum rectangular area to form minimum rectangle encirclement frame, meter in original ginseng image
It calculates the center point coordinate for obtaining minimum rectangle encirclement frame and minimum rectangle is surrounded between the long side and image level direction of frame
Angle theta rotates the long side that original ginseng image makes minimum rectangle surround frame around center point coordinate and is parallel to image level direction,
Frame is surrounded using minimum rectangle to cut original ginseng image, retains the image-region that minimum rectangle is surrounded within the scope of frame,
The image-region that minimum rectangle surrounds the original ginseng image other than frame is deleted, so that ginseng target is full of whole image, most
The result figure such as f in Fig. 8 is obtained eventually.Fig. 9 is the partition strategy that ginseng be uniformly divided into 8 pieces, by the size of ginseng image
Unified 400 × 200 size is zoomed in and out, image uniform is then divided into muti-piece image block, specific implementation divides the image into 8
The big ginseng image block such as block, the size of each piece of image block are 100 × 100 size, calculate ginseng in each image block
Image weights.Figure 10 is to scale it to each piecemeal for 16 × 16 sizes, is then binary map by its binaryzation, piecemeal
The combination order of Hash coding from top to bottom, is composed from left to right, is encoded by the Hash for the pixel value in binary map
The shape of ginseng is indicated.The phase of each piecemeal with the Hash coding of corresponding sub-block in image library is calculated according to Hamming distance
Like degree distance, the similarity distance of all piecemeals of ginseng image is added with the product of the weight of corresponding piecemeal to get having arrived finally
Similarity distance, it is ranked up from small to large, apart from it is the smallest be the most like ginseng image that retrieves.
The present invention may be implemented quickly and accurately to retrieve similar ginseng image, have in field of image search larger
Application value.
Above-mentioned specific embodiment is used to illustrate the present invention, rather than limits the invention, of the invention
In spirit and scope of protection of the claims, to any modifications and changes that the present invention makes, protection model of the invention is both fallen within
It encloses.
Claims (5)
1. a kind of method for ginseng image retrieval, it is characterised in that include the following steps:
1) original ginseng image is the image of RGB color, goes to hsv color space from RGB color to ginseng image,
The filtering of color is carried out to image, image progress binaryzation is then obtained into the first binary map again, by obtained binary map and original
Beginning ginseng image carries out and operation, and obtained image is the ginseng image for removing most of background；
2) the ginseng image of the most of background of removal obtained in step 1) is gone into hsv color from RGB color again
Space is further filtered the pixel lower than hsv color threshold value using hsv color threshold value, then carries out binaryzation and obtains
To the second binary map, largest connected domain is searched in the second binary map, removes other connected domains other than largest connected domain,
Third binary map is obtained, carries out and operates using third binary map as mask and original ginseng image, obtained image is most
Ginseng image after whole background removal；
3) ginseng in the ginseng image for obtaining step 2) must be removed using morphological operation；
4) the ginseng image for the tiny root of removal that step 3) obtains is subjected to rotation correction direction, then ginseng image is carried out
It cuts so that ginseng is full of whole image；
5) the ginseng image in step 4) is subjected to uniform piecemeal, calculates ginseng target in each image block and accounts for whole image
Ratio, obtain the weight of image block, processing obtains Hash coding；Then it is encoded according to Hash and calculates similarity distance, according to
Similarity distance carries out retrieval process.
2. a kind of method for ginseng image retrieval according to claim 1, it is characterised in that:
The color filtering is under the color space of HSV, and the threshold value of setting H, S, V carry out color filtering.
3. a kind of method for ginseng image retrieval according to claim 1, it is characterised in that: the tool of the step 3)
Steps are as follows for body:
3.1) the ginseng image after background removal that step 2) obtains is subjected to binarization operation and obtains the 4th binary map, then carried out
Top cap operation obtains top cap operating result figure, obtains the tiny root part of ginseng in top cap operating result figure；
3.2) top cap operating result figure and the 4th binary map are subjected to xor operation, obtain exclusive or result figure, in exclusive or result figure
For the ginseng binary map for eliminating tiny root；
3.3) the largest connected domain in exclusive or result figure is searched, is the two-value for eliminating the tiny root of ginseng in the maximum UNICOM domain
Largest connected domain and original ginseng image are carried out the ginseng figure that the final tiny root of removal is obtained with operation by image-region
Picture.
4. a kind of method for ginseng image retrieval according to claim 1, it is characterised in that: the tool of the step 4)
Steps are as follows for body:
4.1) gray processing will be carried out through step 3) treated ginseng image first and obtain grayscale image, then grayscale image is subjected to two-value
Change operation and obtains the 5th binary map；
4.2) largest connected region is found in the 5th binary map, an encirclement will be established using the largest connected region as profile
The minimum rectangular area in the largest connected region, and minimum rectangular area is drawn to form minimum rectangle in original ginseng image
Frame is surrounded, minimum rectangle is calculated and surrounds the center point coordinate of frame and long side and the image level side of minimum rectangle encirclement frame
Angle theta between rotates the long side that original ginseng image makes minimum rectangle surround frame around center point coordinate and is parallel to image
Horizontal direction；
4.3) frame is surrounded using minimum rectangle to cut original ginseng image, retain the figure that minimum rectangle is surrounded within the scope of frame
As region, the image-region that minimum rectangle surrounds the original ginseng image other than frame is deleted, so that ginseng target is full of entire
Image.
5. a kind of method for ginseng image retrieval according to claim 1, it is characterised in that: the tool of the step 5)
Steps are as follows for body:
5.1) size of the ginseng image obtained to step 4) zooms in and out unified size, and image uniform is then divided into muti-piece
Image block；
5.2) ginseng image weights in each image block are calculated:
Processing obtains the area d of ginseng part in area S and each image block occupied by ginseng part in whole imagei, i=
1,2 ..., 8, i indicate image block serial number, and then the weighted value that ginseng Partial Feature is extracted in each piece of image block is calculated
Wi=di/ S, the ginseng image weights as each image block；
5.3) each image block is narrowed down into uniform sizes size；Binary conversion treatment is carried out to the image block after diminution, then will
The pixel of ginseng main part is assigned a value of 1, and the pixel in other regions is assigned a value of 0, and building obtains two be made of 0,1 numerical value
Value figure is established the sequence of pixel values being made of 0,1 numerical value and is encoded as the Hash of the image block；
5.4) each image in the image library for inputting ginseng image and inquiry to be retrieved, is all made of aforesaid way acquisition
Image block and its Hash coding；Ginseng image to be retrieved is inputted, each of which image block and inquiry are calculated according to Hamming distance
Similarity distance in image library between the image block of each image, by the similarity distance of all image blocks and correspondence image block
The products of ginseng image weights be added, as final similarity distance, finally according to being ranked up from small to large, distance is minimum
As the most like ginseng image retrieved.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910491812.3A CN110263205B (en) | 2019-06-06 | 2019-06-06 | Retrieval method for ginseng image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910491812.3A CN110263205B (en) | 2019-06-06 | 2019-06-06 | Retrieval method for ginseng image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110263205A true CN110263205A (en) | 2019-09-20 |
| CN110263205B CN110263205B (en) | 2023-07-21 |
Family
ID=67917118
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910491812.3A Active CN110263205B (en) | 2019-06-06 | 2019-06-06 | Retrieval method for ginseng image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110263205B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111626304A (en) * | 2020-05-20 | 2020-09-04 | 中国科学院新疆理化技术研究所 | Color feature extraction method based on machine vision and application thereof |
| CN112115292A (en) * | 2020-09-25 | 2020-12-22 | 海尔优家智能科技（北京）有限公司 | Picture searching method and device, storage medium and electronic device |
| CN113704532A (en) * | 2020-11-25 | 2021-11-26 | 天翼智慧家庭科技有限公司 | Method and system for improving recall rate of picture retrieval |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101403703A (en) * | 2008-11-07 | 2009-04-08 | 清华大学 | Real-time detection method for foreign fiber in lint |
| CN102819582A (en) * | 2012-07-26 | 2012-12-12 | 华数传媒网络有限公司 | Quick searching method for mass images |
| CN103440348A (en) * | 2013-09-16 | 2013-12-11 | 重庆邮电大学 | Vector-quantization-based overall and local color image searching method |
| US20140072219A1 (en) * | 2012-09-08 | 2014-03-13 | Konica Minolta Laboratory U.S.A., Inc. | Document image binarization and segmentation using image phase congruency |
| CN104778450A (en) * | 2015-03-30 | 2015-07-15 | 深圳市道通科技股份有限公司 | Preprocessing method and device for VIN (vehicle identification number) image |
| KR20150117385A (en) * | 2014-04-10 | 2015-10-20 | 삼성전자주식회사 | Image tagging method and apparatus thereof |
| CN107633503A (en) * | 2017-08-01 | 2018-01-26 | 南京航空航天大学 | The image processing method of stalk is remained in a kind of automatic detection grain |
| CN107749057A (en) * | 2017-09-16 | 2018-03-02 | 河北工业大学 | A kind of method of solar battery sheet outward appearance spillage defects detection |
-
2019
- 2019-06-06 CN CN201910491812.3A patent/CN110263205B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101403703A (en) * | 2008-11-07 | 2009-04-08 | 清华大学 | Real-time detection method for foreign fiber in lint |
| CN102819582A (en) * | 2012-07-26 | 2012-12-12 | 华数传媒网络有限公司 | Quick searching method for mass images |
| US20140072219A1 (en) * | 2012-09-08 | 2014-03-13 | Konica Minolta Laboratory U.S.A., Inc. | Document image binarization and segmentation using image phase congruency |
| CN103440348A (en) * | 2013-09-16 | 2013-12-11 | 重庆邮电大学 | Vector-quantization-based overall and local color image searching method |
| KR20150117385A (en) * | 2014-04-10 | 2015-10-20 | 삼성전자주식회사 | Image tagging method and apparatus thereof |
| CN104778450A (en) * | 2015-03-30 | 2015-07-15 | 深圳市道通科技股份有限公司 | Preprocessing method and device for VIN (vehicle identification number) image |
| CN107633503A (en) * | 2017-08-01 | 2018-01-26 | 南京航空航天大学 | The image processing method of stalk is remained in a kind of automatic detection grain |
| CN107749057A (en) * | 2017-09-16 | 2018-03-02 | 河北工业大学 | A kind of method of solar battery sheet outward appearance spillage defects detection |
Non-Patent Citations (1)
| Title |
|---|
| 裴蓓;王朔中;倪丽佳;: "面向基于内容图像检索的图像感知Hash", 上海大学学报(自然科学版) * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111626304A (en) * | 2020-05-20 | 2020-09-04 | 中国科学院新疆理化技术研究所 | Color feature extraction method based on machine vision and application thereof |
| CN111626304B (en) * | 2020-05-20 | 2023-08-04 | 中国科学院新疆理化技术研究所 | Color feature extraction method based on machine vision and application thereof |
| CN112115292A (en) * | 2020-09-25 | 2020-12-22 | 海尔优家智能科技（北京）有限公司 | Picture searching method and device, storage medium and electronic device |
| CN113704532A (en) * | 2020-11-25 | 2021-11-26 | 天翼智慧家庭科技有限公司 | Method and system for improving recall rate of picture retrieval |
| CN113704532B (en) * | 2020-11-25 | 2024-04-26 | 天翼数字生活科技有限公司 | Method and system for improving picture retrieval recall rate |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110263205B (en) | 2023-07-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110263205A (en) | A kind of search method for ginseng image | |
| CN103870516B (en) | Retrieve the method for image, paint in real time reminding method and its device | |
| CN109558908B (en) | Method for determining optimal edge of given area | |
| CN110070091A (en) | The semantic segmentation method and system rebuild based on dynamic interpolation understood for streetscape | |
| CN102819582A (en) | Quick searching method for mass images | |
| WO2022227770A1 (en) | Method for training target object detection model, target object detection method, and device | |
| CN110321769A (en) | A kind of more size commodity on shelf detection methods | |
| CN103413288A (en) | LCD general defect detecting method | |
| CN114299066B (en) | Defect detection method and device based on salient feature pre-extraction and image segmentation | |
| CN104036493B (en) | No-reference image quality evaluation method based on multifractal spectrum | |
| CN102521600A (en) | Method and system for identifying white-leg shrimp disease on basis of machine vision | |
| CN105184225B (en) | A kind of multinational banknote image recognition methods and device | |
| CN102779157A (en) | Method and device for searching images | |
| CN110427792A (en) | A kind of recognition matching process blocking two dimensional code and its application | |
| CN115456956A (en) | Method and device for detecting scratches of liquid crystal display and storage medium | |
| CN103399863B (en) | Image search method based on the poor characteristic bag of edge direction | |
| CN104463091B (en) | A kind of facial image recognition method based on image LGBP feature subvectors | |
| CN115240119A (en) | Pedestrian small target detection method in video monitoring based on deep learning | |
| CN114092343A (en) | Cloud removing method for remote sensing image | |
| CN103455816B (en) | Stroke width extraction method and device and character recognition method and system | |
| CN115731257A (en) | Leaf form information extraction method based on image | |
| CN114612709A (en) | Multi-scale target detection method guided by image pyramid characteristics | |
| CN110490250A (en) | A kind of acquisition methods and device of artificial intelligence training set | |
| CN107451604A (en) | A kind of image classification method based on K means | |
| Shen et al. | Surface defect detection algorithm of hot-rolled strip based on improved YOLOv7 |
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 |