CN104636754A - Intelligent image classifying method based on tongue body partition color feature - Google Patents
Intelligent image classifying method based on tongue body partition color feature Download PDFInfo
- Publication number
- CN104636754A CN104636754A CN201510059796.2A CN201510059796A CN104636754A CN 104636754 A CN104636754 A CN 104636754A CN 201510059796 A CN201510059796 A CN 201510059796A CN 104636754 A CN104636754 A CN 104636754A
- Authority
- CN
- China
- Prior art keywords
- tongue body
- tongue
- region
- picture
- image
- 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
Abstract
The invention provides an intelligent image classifying method based on a tongue body partition color feature. The method comprises the following steps: training a large amount of tongue body picture data based on a histogram of oriented gradients feature, and positioning a tongue body in a video by using a training result; performing more accurate segmentation; partitioning a segmented tongue body region into different regions, extracting the pixels of each region for performing separation on an RGB (Red Green Blue) color space, and counting color histograms in each region; performing similarity computation according to the color histograms and regional positions, and outputting a tongue body picture which is most similar to a video tongue body region. The detection accuracy is high, and a picture in which different regions are similar in colors can be found by means of tongue body region partition. An output video image classifying result can be applied to identity authentication based on tongue body features as an auxiliary identification means in multi-modal identity verification, and can be applied to biological information image classification in artificial intelligent application.
Description
Technical field
The present invention relates to the video image of artificial intelligence and the Automatic Measurement Technique field of machine vision, be specifically related to the intelligent image sorting technique based on tongue body subregion color characteristic.
Background technology
At present, the fields such as national security, finance and ecommerce are easily subject to the serious threat palming off identity person, and the importance of identification becomes increasingly conspicuous.In recent years, the smart identity authentication technique based on living things feature recognition more and more receives publicity.Living things feature recognition be utilize image procossing, the method for pattern-recognition and computer vision effectively analyzes the physiology of the uniqueness that human body has and behavioural characteristic and describes, and by judge these analyze and description consistance and confirm a class technology of identity.Method generally based on physiological characteristic has fingerprint recognition, iris recognition, recognition of face, hand-type identification etc.Through the confirmation of sample view and medical expert, contain a large amount of features varied with each individual in mankind's tongue body, the outward appearance of tongue body varies, and is mainly reflected in the color of tongue body, on the groove mark on shape and surface and grain distribution, after growing up, the change of tongue body is very small.Thus, the tongue body of people can as a new biological characteristic mode, for auxiliary multi-modal identity information identification.
Simultaneously due to the development of network and multimedia technology, people create a large amount of graph datas, the explosive growth mode that digital picture presents in work and daily life.Comprise the medical image of tongue body image, scanned picture, the bulk information of news picture etc. is submerged in various data, if do not have effective image data management mode, the picture needed cannot be judged and be classified, cannot be retrieved out when needed.By the Image Classfication Technology of artificial intelligence, tissue typing can be carried out to the mass image data of a certain class image (the tongue body view data as in this method), and find similar picture in this, as user, store, thumbnail data are very important.At present, how efficiently the feature of combining image, as color characteristic, textural characteristics etc., carry out quantitative test to image, according to the brightness of image, tone, position and architectural feature, make to image scene type or target the research contents that correct judgement and explanation are focuses exactly.
Still there is a lot of problem in existing tongue body image mechanized classification research.Current existing method is difficult to split from image accurately, in robustness, accuracy and practicality all can not meet the needs in subsequent biological graphical analysis.Relatively not accurate segmentation result, the analysis that also can affect the later stage is to a great extent sorted out.Further, the tongue body graphical analysis classifying method of Most current is only according to some independently features, and as color carries out identifying and analyzing, lack in individuality the analytical approach combined.The present invention can accurate allocation tongue body region, and utilizes watershed algorithm to carry out the segmentation of high-accuracy to tongue body.For result after segmentation, tongue body is divided into zones of different by the present invention, and the color characteristic according to zones of different makes classification to picture, and can judge to find out the most close picture on the basis of calmodulin binding domain CaM and color characteristic.Can be applied in following multi-modal identification system based on the biological information picture classification of tongue characteristic and the identification of tongue body information.
Summary of the invention
The present invention is directed to the deficiency of existing tongue body Images Classification automation research method, provide the intelligent image sorting technique based on tongue body subregion color characteristic.The object of the present invention is to provide robotization and the tongue body image find out the method for similar image of can classifying in high accuracy, can be used for multi-modal biological characteristic identification, tongue characteristic is utilized to carry out identity information identification as aid identification means, the biological information Images Classification in artificial intelligence can also be used for, reduce the error that subjective factor affects and artificial judgment is brought.
Based on the intelligent image sorting technique of tongue body subregion color characteristic, comprise the following steps:
A () is normalized the tongue body image gathered, extract its histograms of oriented gradients feature and train;
B () is classified based on the training result in SVM classifier team's step (a), locate and mark the rectangle band of position of tongue body;
C (), for the video data of input, utilizes the tongue body band of position that the principle in step (b) obtains, and carries out tongue body segmentation by watershed algorithm;
D (), to obtain the rectangular area of tongue body location in step (b), the result according to tongue body segmentation adjusts rectangular area, make it comprise all parts of tongue body segmentation.The rectangular area of the tongue body again obtained after adjustment location is divided into N number of subregion;
E () N number of subregion to step (d) extracts color characteristic respectively and calculates its color histogram;
F histogram data corresponding with picture library respectively for respective color histogram in step (e) is done similarity comparison by (), the similarity result in the N number of region calculated is got average and read the highest picture of similarity.
In the step (a) of the above-mentioned intelligent image sorting technique based on tongue body subregion color characteristic, training method is, tongue body picture is divided into positive sample and negative sample, wherein positive 52, sample, and negative sample 183 is put respectively to two files.The sample obtained is carried out to the adjustment of size, size is 256*320.After all image zoomings to uniform sizes, extract the histograms of oriented gradients feature of all positive samples, afterwards the same histograms of oriented gradients feature extracting all negative samples.
In the step (a) of the above-mentioned intelligent image sorting technique based on tongue body subregion color characteristic, the computing method of direction gradient figure are, several blocks are divided into for every width image, zoning is continued to each block, to each cell unit of last division, calculated direction histogram of gradients.Direction gradient is by the horizontal direction gradient G of pixel (x, y)
x(x, y)=H (x+1, y)-H (x-1, y) and vertical direction gradient G
y(x, y)=H (x, y+1)-H (x, y-1) calculates respectively, and wherein H (x, y) represents pixel value, and by
calculate gradient magnitude, by
Calculate gradient direction.
In the step (b) of the above-mentioned intelligent image sorting technique based on tongue body subregion color characteristic, tongue body localization method is, is L with wide, and height is the detection window of H, and wherein L and H is preset value; In step (c), obtain input picture, detection window according to the step-length STEP of setting from left to right, slides from top to bottom, and obtains the higher-dimension direction gradient descriptor vector of detection window, and then obtains the distance of Optimal Separating Hyperplane.Thus, SVM classifier is utilized to obtain the rectangle positional information of the tongue body obtained in input picture.
In the step (c) of the above-mentioned intelligent image sorting technique based on tongue body subregion color characteristic, the method of tongue body segmentation is, for the video frame images of input, by the tongue body rectangle band of position obtained in step (b), its top left corner apex coordinate is (x
l, y
l), long is L, and height is H.Make a mark to tongue body region afterwards, the elliptic region of mark is the center (x with rectangular area
l+ 0.5L, y
l+ 0.5H) be elliptical center, take 3/7L as short radius, 3/7H is major radius.Value according to mark utilizes watershed algorithm to make segmentation to the tongue body region in image.
In the step (d) of the above-mentioned intelligent image sorting technique based on tongue body subregion color characteristic, for carrying out the picture after dividing processing in step (c), by the acquisition in step (b) with top left corner apex coordinate for (x
l, y
l), long is L, and height is H tongue body locating rectangle region.To growing for L, height adjusts for H tongue body locating rectangle region.Judge rectangular area whether comprise all segmentations after tongue body region, if the size constancy of then rectangular area, if not then adjust the wide and high of rectangle, make it just comprise the tongue body region of segmentation; Rectangular area after adjustment is top left corner apex coordinate is (x, y), and long is L
n, height is H
nby the image in region with center (x+0.5L
n, y+0.5H
n) be divided into N block (N can set in software).
In the step (e) of the above-mentioned intelligent image sorting technique based on tongue body subregion color characteristic, for N number of rectangular area of decile in step (d), each rectangular area is extracted in RGB color space to tongue body pixel after the segmentation in respective regions, and preserve with the form of corresponding N number of matrix image.Respectively R is gone out to N number of separation of images, G, B tri-Color Channels, and then calculate its color histogram, the wherein element number B of color histogram
bins, G
bins, R
binsbe set to 16 respectively, also can be preset as 32 etc. according to effect.For the result drawn in compute histograms, find out histogrammic maximal value V
max, again travel through histogram afterwards, by all values divided by V
maxbe normalized.The scope of the value of each unit after normalization is [0,1].
In the step (f) of the above-mentioned intelligent image sorting technique based on tongue body subregion color characteristic, all pictures of picture library are classified according to tongue body color, is respectively pink tongue, deep red red tongue, yellow tongue tongue, blue-purple tongue, white tongue tongue, grey and black coat tongue, totally 6 types.
In the step (f) of the above-mentioned intelligent image sorting technique based on tongue body subregion color characteristic, after the picture segmentation in picture library, extract color histogram data according to step (d) and step (e).The histogram data in N number of region of the frame picture of input video is calculated similarity respectively with corresponding N number of region of picture in picture library respectively, and averaging to the similarity in N number of region obtains V
sim.According to the similarity V obtained in classification chart valut
sim, in picture library, choose one that similarity is the highest, and then classification made to tongue body image in video.
Intelligent image sorting technique based on tongue body subregion color characteristic provided by the invention, first, based on histograms of oriented gradients feature, normalized tongue body view data is trained, the video data of SVM classifier to input is utilized to classify, then be L by length, the tongue body region that the detection window that height is H is oriented, afterwards according to the region of location, utilizes watershed algorithm to split more accurately.Then the tongue body region be partitioned into is adjusted, to adjust center, rear rectangular area to be divided into different regions, the pixel extracting each region is separated on color space, to each range statistics color histogram, the last calculating doing similarity according to color histogram and regional location, exports the picture the most similar to video tongue body region and corresponding pathological characters.Meanwhile, the present invention can the different pathological characters picture library of support maintenance.
Compared with prior art, tool of the present invention has the following advantages and technique effect:
Previous work based on the intelligent image sorting technique of tongue characteristic is the region segmentation to tongue body, if do not have good tongue body segmentation effect, follow-up work is just difficult to carry out.Tongue body dividing method in the past adopts artificial segmentation, and existing method mainly contains the dividing method based on region growing, based on the maximum variance between clusters of OUTS, and edge detection method, color space clustering procedure etc., and often kind of method has its limitation.Such as, the dividing method based on region growing is difficult to choose for similar value, is easily chosen in the region such as lip and tooth; Maximum variance between clusters for not obviously bimodal or ambient brightness impact is large time be difficult to good volume segmentation effect.The present invention adopts the tongue body telltale mark dividing method based on watershed algorithm, can obtain high robust, the result of high-accuracy to the tongue body segmentation of the two field picture of the video in practical application.Existing tongue body image classification method is mostly only classified according to single feature, and the result of classifying can be caused only similar in a certain feature, but it seems that aggregate analysis can not be classified as a class.This method is classified based on the subregion color characteristic of tongue body, approximate picture is chosen when provincial characteristics is similar, can be applicable to the identification based on tongue characteristic in multi-modal biological information identification, as the aid identification means in multi-modal identification system, also can be applicable to intelligent classification and the search of the biological information image in artificial intelligence.
Accompanying drawing explanation
Fig. 1 is the intelligent image sorting technique based on tongue body subregion color characteristic in embodiment.
Embodiment
Be described further embodiments of the present invention below in conjunction with embodiment, but enforcement of the present invention being not limited thereto, not describing part in detail especially if having below, is all that those skilled in the art can refer to existing techniques in realizing.
As Fig. 1, the intelligent image sorting technique based on tongue characteristic comprises the following steps:
The first step, positions the tongue body region in video.
Data training method before location is as follows: tongue body picture is divided into positive sample and negative sample, wherein positive 52, sample, and negative sample 183 is put respectively to two files.The sample obtained is carried out to the adjustment of size, size is 256*320.After all image zoomings to uniform sizes, extract the histograms of oriented gradients feature of all positive samples, afterwards the same histograms of oriented gradients feature extracting all negative samples.The computing method of histograms of oriented gradients are as follows: every width image is divided into several blocks, divides some subregions to each block, to each cell unit of last division, and calculated direction histogram of gradients.Direction gradient is by the horizontal direction gradient G of pixel (x, y)
x(x, y)=H (x+1, y)-H (x-1, y) and vertical direction gradient G
y(x, y)=H (x, y+1)-H (x, y-1) calculates respectively, and wherein H (x, y) represents pixel value, and by
calculate gradient magnitude, by
calculate gradient direction.Afterwards 1 and 0 are labeled as respectively for positive sample and negative sample, its histogram of gradients characteristic sum label value is trained with SVM respectively.
Given size is adjusted to the frame picture of the video of input, utilize SVM classifier to do according to the data trained based on histograms of oriented gradients to classify, and then set wide for L, height is the detection window of H, make its step-length STEP according to the rules from left to right, carry out from top to bottom sliding and location is made to tongue body, make a mark with rectangular area.
Second step, splits tongue body region.
For the two field picture of the video of input, by the tongue body rectangle band of position obtained in the first step, its top left corner apex coordinate is (x
l, y
l), wide is L, and height is H.Afterwards with the center (x of rectangular area
l+ 0.5L, y
l+ 0.5H) be elliptical center, take 3/7L as short radius, 3/7H is major radius beginning label tongue body region.Watershed algorithm is utilized to make segmentation to the tongue body region in image according to the mark made.
3rd step, carries out subregion to tongue body region.
What obtain in previous step is (x with top left corner apex coordinate
l, y
l), long is L, and height is H tongue body locating rectangle region.To growing for L, height adjusts for H tongue body locating rectangle region.Judge rectangular area whether comprise all segmentations after tongue body region, if the size constancy of then rectangular area, if not then adjust the wide and high of rectangle, make it just comprise the tongue body region of segmentation.Rectangular area after adjustment is top left corner apex coordinate is (x, y), and long is L
n, height is H
nby the image in region with center (x+0.5L
n, y+0.5H
n) be divided into N block.
4th step, extracts the color characteristic of subregion after subregion respectively.
N number of rectangular area of decile in the third step, extracts tongue body pixel after the segmentation in respective regions, and preserves with the form of corresponding N number of matrix image in RGB color space to each rectangular area.Respectively R is gone out to this N number of separation of images, G, B tri-Color Channels, and then calculate its color histogram, the element number B of color histogram
bins, G
bins, R
binsbe set to 16 respectively.The result drawn in compute histograms, finds out histogrammic maximal value V
max, again travel through histogram afterwards, by all values divided by V
maxbe normalized.The scope of the value of each unit after normalization is [0,1], just obtains afterwards comprising R, the subregion color histogram of G, B tri-Color Channels.
5th step, calculates picture similarity.
Color histogram data is extracted according to the 4th step after picture segmentation in picture library.The histogram data in N number of region of the frame picture of input video is calculated similarity respectively with corresponding N number of region of picture in picture library respectively, and averaging to the similarity in this N number of region obtains V
sim.According to the similarity V obtained in classification chart valut
sim, in picture library, choose one that similarity is the highest, and then classification made to tongue body picture in video.
Claims (10)
1., based on the intelligent image sorting technique of tongue body subregion color characteristic, it is characterized in that, comprise the following steps:
A () is normalized for the tongue body image of camera collection, extract its histograms of oriented gradients feature, and the histograms of oriented gradients characteristic according to extracting is trained;
B (), based on SVM classifier, is classified according to the training result in step (a), position thus to tongue body region, obtains the rectangle positional information of tongue body;
C () input, with the video data of tongue body, is utilized the rectangle positional information of the tongue body obtained in step (b), is split by watershed algorithm to tongue body;
D (), with the rectangle band of position of the tongue body obtained in step (b), the result according to the segmentation of step (c) tongue body adjusts rectangular area, make it comprise all parts of tongue body segmentation; Rectangular area after adjustment is divided into N number of rectangular area, and N can be 4,6 or 9;
E () extracts color characteristic respectively for N number of region of step (d) decile, and calculate its color histogram;
The classification chart valut of (f) Criterion tongue body image, the picture for classification chart valut extracts color characteristic according to step (a) ~ (e) equally, and calculates color histogram;
G the histogram data that respective color histogram in step (e) is corresponding with the classification chart valut in step (f) is respectively made Similarity Measure by (), the similarity result in the N number of region calculated is got average, according to the similarity V obtained in classification chart valut
sim, in classification chart valut, choose one that similarity is the highest, this pictures is a tongue body picture the most similar to tongue body in the camera data inputted in step (c).
2. according to claim 1 based on the intelligent image sorting technique of tongue body subregion color characteristic, it is characterized in that: the method for the described training in step (a) is, tongue body picture is divided into positive sample and negative sample, wherein positive 52, sample, negative sample 183 is put to two files respectively; The sample obtained is carried out to the adjustment of size, size is 256*320; After all image zoomings to uniform sizes, extract the histograms of oriented gradients feature of all positive samples, afterwards the same histograms of oriented gradients feature extracting all negative samples.
3. according to claim 1 based on the intelligent image sorting technique of tongue body subregion color characteristic, it is characterized in that: the method for the extraction histograms of oriented gradients in step (a) is: for growing for winWidth, wide winHeight is that every width image is divided into m block, the computing formula of m is m=[(winWidth ?16)/8+1] * [(winHeight ?16)/8+1], the size of each piece is 16*16, namely the parameter blockSize in OpenCv is set to blockSize (16, 16), continue to be divided into 4 regions to each block, each zonule that last division obtains is called a cell element, the parameter cellSize of cell element is set to cellSize (8, 8), in each cell element, gradient direction is divided into 9 intervals, so just obtain the proper vector of one 9 dimension, and then each block has 36 proper vectors, the block be made up of 4 cell elements scans image, the step-length STEP of scanning is 8 pixels, finally the feature of blocks all in image is together in series, just obtain the direction gradient feature of image, wherein, the method for calculated direction histogram of gradients is as follows: with the summit in the image upper left corner for initial point, the limit of horizontal direction is x-axis, the limit of vertical direction is y-axis, and direction gradient is by the horizontal direction gradient G of pixel (x, y)
x(x, y)=H (x+1, y)-H (x-1, y) and vertical direction gradient G
y(x, y)=H (x, y+1)-H (x, y-1) calculates respectively, wherein H (x, y) represents pixel value, and by
calculate gradient magnitude, by
calculate gradient direction.
4. according to claim 1 based on the intelligent image sorting technique of tongue body subregion color characteristic, it is characterized in that: in step (b), tongue body localization method is, is L with wide, and height is the detection window of H, and wherein L and H is preset value; Input picture is obtained in step (c), detection window according to the step-length STEP of setting from left to right, slide from top to bottom, and obtain the higher-dimension direction gradient descriptor vector of detection window, and then obtain the distance of Optimal Separating Hyperplane, thus, SVM classifier is utilized to obtain the rectangle positional information of the tongue body obtained in input picture.
5. according to claim 3 based on the intelligent image sorting technique of tongue body subregion color characteristic, it is characterized in that: in step (c), for the video frame images of input, by the tongue body rectangle band of position obtained in step (b), its top left corner apex coordinate is (x
l, y
l), long is L, and height is H; Make a mark to tongue body region afterwards, the elliptic region of mark is the center (x with rectangular area
l+ 0.5L, y
l+ 0.5H) be elliptical center, take 3/7L as short radius, 3/7H is major radius; Value according to mark utilizes watershed algorithm to make segmentation to the tongue body region in image.
6. according to claim 5 based on the intelligent image sorting technique of tongue body subregion color characteristic, it is characterized in that: in step (d), for carrying out the picture after dividing processing in step (c), by the acquisition in step (b) with top left corner apex coordinate for (x
l, y
l), long is L, and height is H tongue body locating rectangle region.
7. according to claim 5 based on the intelligent image sorting technique of tongue body subregion color characteristic, it is characterized in that: step (d) is L to length, and height adjusts for H tongue body locating rectangle region; Judge rectangular area whether comprise all segmentations after tongue body region, if the size constancy of then rectangular area, if not then adjust the wide and high of rectangle, make it just comprise the tongue body region of segmentation; Rectangular area after adjustment is top left corner apex coordinate is (x, y), and long is L
n, height is H
nby the image in region with center (x+0.5L
n, y+0.5H
n) being divided into N block, N can be set as 4,6 or 9.
8. according to claim 1 based on the intelligent image sorting technique of tongue body subregion color characteristic, it is characterized in that: in step (e), for N number of rectangular area of decile in step (d), each rectangular area is extracted in RGB color space to tongue body pixel after the segmentation in respective regions, and preserve with the form of corresponding N number of matrix image; Respectively R is gone out to N number of separation of images, G, B tri-Color Channels, and then calculate its color histogram; For the result drawn in compute histograms, find out histogrammic maximal value V
max, again travel through histogram afterwards, by all values divided by V
maxbe normalized; The scope of the value of each unit after normalization is [0,1].
9. according to claim 1 based on the intelligent image sorting technique of tongue body subregion color characteristic, it is characterized in that: in step (f), all pictures of picture library are classified according to tongue body color, be respectively pink tongue, deep red red tongue, yellow tongue tongue, blue-purple tongue, white tongue tongue, grey and black coat tongue, totally 6 types.
10. according to claim 1 based on the intelligent image sorting technique of tongue body subregion color characteristic, it is characterized in that: in step (f), after the picture segmentation in picture library, extract color histogram data according to step (d) and step (e); The histogram data in N number of region of input video frame picture is calculated similarity respectively with corresponding N number of region of picture in picture library respectively, and the similarity in N number of region is averaged obtains V
sim.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510059796.2A CN104636754B (en) | 2015-01-31 | 2015-01-31 | Intelligent image sorting technique based on tongue body subregion color characteristic |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510059796.2A CN104636754B (en) | 2015-01-31 | 2015-01-31 | Intelligent image sorting technique based on tongue body subregion color characteristic |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104636754A true CN104636754A (en) | 2015-05-20 |
| CN104636754B CN104636754B (en) | 2018-02-27 |
Family
ID=53215482
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510059796.2A Active CN104636754B (en) | 2015-01-31 | 2015-01-31 | Intelligent image sorting technique based on tongue body subregion color characteristic |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104636754B (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105354576A (en) * | 2015-09-30 | 2016-02-24 | 上海圣尧智能科技有限公司 | Target characteristic extraction method, target characteristic extraction module, target module creating module and intelligent image monitoring apparatus |
| CN106778860A (en) * | 2016-12-12 | 2017-05-31 | 中国矿业大学 | Image position method based on Histogram Matching |
| CN108289793A (en) * | 2015-10-15 | 2018-07-17 | 东森塔克斯公司 | Dosage form distributor based on image recognition |
| CN111310851A (en) * | 2020-03-03 | 2020-06-19 | 四川大学华西第二医院 | Artificial intelligence ultrasonic auxiliary system and application thereof |
| CN111815733A (en) * | 2020-08-07 | 2020-10-23 | 深兰科技（上海）有限公司 | Video coloring method and system |
| CN113256557A (en) * | 2021-04-07 | 2021-08-13 | 北京联世科技有限公司 | Traditional Chinese medicine tongue state identification method and device based on tongue manifestation clinical symptom image |
| WO2022110121A1 (en) * | 2020-11-30 | 2022-06-02 | Orange | Method for biometric analysis about a user of an electronic device |
| CN117197139A (en) * | 2023-11-07 | 2023-12-08 | 天津市肿瘤医院（天津医科大学肿瘤医院） | Tongue diagnosis image multi-label classification method based on AI |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040151379A1 (en) * | 2002-12-28 | 2004-08-05 | Samsung Electronics Co., Ltd. | Method of digital image analysis for isolating a region of interest within a tongue image and health monitoring method and apparatus using the tongue image |
| CN1920854A (en) * | 2006-09-15 | 2007-02-28 | 清华大学 | Tongue division and extracting method for tongue diagnosis used colored digital photo of tongue in Chinese traditional medicine |
| CN101576913A (en) * | 2009-06-12 | 2009-11-11 | 中国科学技术大学 | Automatic clustering, visual and retrieval system for tongue picture based on self-organizing map neural network |
| CN102426583A (en) * | 2011-10-10 | 2012-04-25 | 北京工业大学 | Chinese medicine tongue manifestation retrieval method based on image content analysis |
-
2015
- 2015-01-31 CN CN201510059796.2A patent/CN104636754B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040151379A1 (en) * | 2002-12-28 | 2004-08-05 | Samsung Electronics Co., Ltd. | Method of digital image analysis for isolating a region of interest within a tongue image and health monitoring method and apparatus using the tongue image |
| CN1920854A (en) * | 2006-09-15 | 2007-02-28 | 清华大学 | Tongue division and extracting method for tongue diagnosis used colored digital photo of tongue in Chinese traditional medicine |
| CN101576913A (en) * | 2009-06-12 | 2009-11-11 | 中国科学技术大学 | Automatic clustering, visual and retrieval system for tongue picture based on self-organizing map neural network |
| CN102426583A (en) * | 2011-10-10 | 2012-04-25 | 北京工业大学 | Chinese medicine tongue manifestation retrieval method based on image content analysis |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105354576A (en) * | 2015-09-30 | 2016-02-24 | 上海圣尧智能科技有限公司 | Target characteristic extraction method, target characteristic extraction module, target module creating module and intelligent image monitoring apparatus |
| CN105354576B (en) * | 2015-09-30 | 2019-06-11 | 上海圣尧智能科技有限公司 | A kind of method of target's feature-extraction, target's feature-extraction module, object module creation module and intelligent image monitoring device |
| CN108289793A (en) * | 2015-10-15 | 2018-07-17 | 东森塔克斯公司 | Dosage form distributor based on image recognition |
| CN106778860A (en) * | 2016-12-12 | 2017-05-31 | 中国矿业大学 | Image position method based on Histogram Matching |
| CN111310851A (en) * | 2020-03-03 | 2020-06-19 | 四川大学华西第二医院 | Artificial intelligence ultrasonic auxiliary system and application thereof |
| CN111815733A (en) * | 2020-08-07 | 2020-10-23 | 深兰科技（上海）有限公司 | Video coloring method and system |
| WO2022110121A1 (en) * | 2020-11-30 | 2022-06-02 | Orange | Method for biometric analysis about a user of an electronic device |
| CN113256557A (en) * | 2021-04-07 | 2021-08-13 | 北京联世科技有限公司 | Traditional Chinese medicine tongue state identification method and device based on tongue manifestation clinical symptom image |
| CN117197139A (en) * | 2023-11-07 | 2023-12-08 | 天津市肿瘤医院（天津医科大学肿瘤医院） | Tongue diagnosis image multi-label classification method based on AI |
| CN117197139B (en) * | 2023-11-07 | 2024-02-02 | 天津市肿瘤医院（天津医科大学肿瘤医院） | Tongue diagnosis image multi-label classification method based on AI |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104636754B (en) | 2018-02-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104636754A (en) | Intelligent image classifying method based on tongue body partition color feature | |
| Wei et al. | Multi-vehicle detection algorithm through combining Harr and HOG features | |
| US10943346B2 (en) | Multi-sample whole slide image processing in digital pathology via multi-resolution registration and machine learning | |
| Liu et al. | Detection of aphids in wheat fields using a computer vision technique | |
| CN106651872B (en) | Pavement crack identification method and system based on Prewitt operator | |
| CN107610114B (en) | optical satellite remote sensing image cloud and snow fog detection method based on support vector machine | |
| EP2523165B1 (en) | Image processing method and image processing device | |
| CN102426649B (en) | Simple steel seal digital automatic identification method with high accuracy rate | |
| CN104268583B (en) | Pedestrian re-recognition method and system based on color area features | |
| EP2833288B1 (en) | Face calibration method and system, and computer storage medium | |
| CN102194108B (en) | Smile face expression recognition method based on clustering linear discriminant analysis of feature selection | |
| CN107392968B (en) | The image significance detection method of Fusion of Color comparison diagram and Color-spatial distribution figure | |
| CN103996046B (en) | The personal identification method merged based on many visual signatures | |
| CN107909081B (en) | Method for quickly acquiring and quickly calibrating image data set in deep learning | |
| CN109918971B (en) | Method and device for detecting number of people in monitoring video | |
| CN103971126A (en) | Method and device for identifying traffic signs | |
| CN108629783A (en) | Image partition method, system and medium based on the search of characteristics of image density peaks | |
| Liang et al. | Salient object detection using content-sensitive hypergraph representation and partitioning | |
| CN110956099B (en) | Dynamic gesture instruction identification method | |
| CN102542244A (en) | Face detection method and system and computer program product | |
| CN107590500A (en) | A kind of color recognizing for vehicle id method and device based on color projection classification | |
| Galsgaard et al. | Circular hough transform and local circularity measure for weight estimation of a graph-cut based wood stack measurement | |
| CN106909883A (en) | A kind of modularization hand region detection method and device based on ROS | |
| CN106909884A (en) | A kind of hand region detection method and device based on hierarchy and deformable part sub-model | |
| Dargham et al. | Lips detection in the normalised RGB colour scheme |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |