CN108389207A - A kind of the tooth disease diagnosing method, diagnostic device and intelligent image harvester - Google Patents
A kind of the tooth disease diagnosing method, diagnostic device and intelligent image harvester Download PDFInfo
- Publication number
- CN108389207A CN108389207A CN201810403336.0A CN201810403336A CN108389207A CN 108389207 A CN108389207 A CN 108389207A CN 201810403336 A CN201810403336 A CN 201810403336A CN 108389207 A CN108389207 A CN 108389207A
- Authority
- CN
- China
- Prior art keywords
- odontopathy
- image
- tooth
- detected
- tooth regions
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 55
- 208000010641 Tooth disease Diseases 0.000 title claims abstract description 51
- 238000012545 processing Methods 0.000 claims abstract description 13
- 238000003384 imaging method Methods 0.000 claims description 30
- 238000013527 convolutional neural network Methods 0.000 claims description 25
- 238000004422 calculation algorithm Methods 0.000 claims description 15
- 238000001514 detection method Methods 0.000 claims description 14
- 238000007789 sealing Methods 0.000 claims description 14
- 238000012549 training Methods 0.000 claims description 14
- 238000011156 evaluation Methods 0.000 claims description 10
- 238000004891 communication Methods 0.000 claims description 9
- 239000012528 membrane Substances 0.000 claims description 8
- 230000008569 process Effects 0.000 claims description 8
- 230000002708 enhancing effect Effects 0.000 claims description 7
- 238000001914 filtration Methods 0.000 claims description 2
- 235000013399 edible fruits Nutrition 0.000 claims 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims 1
- 238000003745 diagnosis Methods 0.000 abstract description 17
- 238000010801 machine learning Methods 0.000 abstract description 6
- 238000013135 deep learning Methods 0.000 abstract description 5
- 238000002405 diagnostic procedure Methods 0.000 abstract description 4
- 238000010586 diagram Methods 0.000 description 7
- 238000013507 mapping Methods 0.000 description 7
- 230000011218 segmentation Effects 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 238000009825 accumulation Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000005286 illumination Methods 0.000 description 3
- 238000007726 management method Methods 0.000 description 3
- 238000005259 measurement Methods 0.000 description 3
- 241001124569 Lycaenidae Species 0.000 description 2
- 230000009471 action Effects 0.000 description 2
- 239000013013 elastic material Substances 0.000 description 2
- 230000005611 electricity Effects 0.000 description 2
- 210000001525 retina Anatomy 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 208000018035 Dental disease Diseases 0.000 description 1
- VYPSYNLAJGMNEJ-UHFFFAOYSA-N Silicium dioxide Chemical compound O=[Si]=O VYPSYNLAJGMNEJ-UHFFFAOYSA-N 0.000 description 1
- 230000002547 anomalous effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 229910002056 binary alloy Inorganic materials 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 229920002457 flexible plastic Polymers 0.000 description 1
- 238000007667 floating Methods 0.000 description 1
- 230000003902 lesion Effects 0.000 description 1
- 230000000873 masking effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 210000000214 mouth Anatomy 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 238000005375 photometry Methods 0.000 description 1
- 239000002985 plastic film Substances 0.000 description 1
- 238000004540 process dynamic Methods 0.000 description 1
- 238000001303 quality assessment method Methods 0.000 description 1
- 238000013441 quality evaluation Methods 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 210000005181 root of the tongue Anatomy 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 239000000741 silica gel Substances 0.000 description 1
- 229910002027 silica gel Inorganic materials 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000013316 zoning Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/285—Selection of pattern recognition techniques, e.g. of classifiers in a multi-classifier system
-
- G06T5/94—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
- G06T7/75—Determining position or orientation of objects or cameras using feature-based methods involving models
-
- 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
- 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
- G06V10/758—Involving statistics of pixels or of feature values, e.g. histogram matching
-
- 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
- G06T2207/10004—Still image; Photographic image
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30036—Dental; Teeth
Abstract
The present invention discloses a kind of the tooth disease diagnosing method, diagnostic device and intelligent image harvester, belongs to technical field of image processing.A kind of the tooth disease diagnosing method of the present invention, including obtain tooth regions image to be detected；The tooth regions image to be detected is input to odontopathy identification model and carries out odontopathy identification, and exports odontopathy recognition result, the odontopathy recognition result includes odontopathy position and odontopathy type.A kind of the tooth disease diagnosing method, diagnostic device and the intelligent image harvester of the present invention is by using the image dental diagnostic method based on deep learning, the automatic recognition classification method that the diagnosis method of discrimination that protracted experience is accumulated will be needed to be converted into machine learning in conventional dental, the efficiency for effectively improving auxiliary diagnosis realizes the automation of the tooth disease diagnosing.
Description
Technical field
The present invention relates to a kind of the tooth disease diagnosing method, diagnostic device and intelligent image harvesters, belong to image procossing neck
Area of pattern recognition in domain.
Background technology
China always exists the problem of the difficulty of getting medical service at present, is mainly reflected in that patient is more, and doctor is few, sees a doctor Shi doctor's needs
It inquires, diagnose for each patient, see a doctor inefficient.The tooth disease diagnosing is the exception occurred according to tooth, gingival position
The method that lesion characteristics carry out auxiliary diagnosis often can be in tooth according to dentist it was verified that when the tooth of people is there are when illness
There are the anomalous variations such as corresponding form, color, texture in tooth or the different position of gum, and it is special that traditional teeth disease can be based on these
The presentation of sign variation diagnoses dental disorders, but the personal experience that these diagnosis are based on doctor judges therefore, how to improve
The tooth disease diagnosing efficiency realizes that the tooth disease diagnosing of automation is always urgent problem.
Invention content
The purpose of the present invention is to provide a kind of the tooth disease diagnosing method, diagnostic device and intelligent image harvesters, pass through
Tooth regions image to be detected is obtained, and it is input to odontopathy identification model progress odontopathy and identifies to obtain odontopathy position and odontopathy
Type, intelligence is convenient to realize that the diagnosis of odontopathy will be in conventional dental using the image dental diagnostic method based on deep learning
It needs the diagnosis method of discrimination that protracted experience is accumulated to be converted into the automatic recognition classification method of machine learning, effectively improves auxiliary
The efficiency of diagnosis realizes the automation of the tooth disease diagnosing.
It is as follows that the present invention provides technical solution：
On the one hand, the present invention provides a kind of the tooth disease diagnosing methods, including：
Obtain tooth regions image to be detected；
The tooth regions image to be detected is input to odontopathy identification model and carries out odontopathy identification, and exports odontopathy knowledge
Not as a result, the odontopathy recognition result includes odontopathy position and odontopathy type.
According to an embodiment of the present invention, described the step of obtaining tooth regions image to be detected, includes：
Image to be detected is acquired using intelligent image harvester；
Described image to be detected is detected based on AdaBoost detection algorithms and whether there is tooth regions, if so, judgement institute
It is tooth regions image to be detected to state image to be detected.
Another embodiment according to the present invention, it is described to input the tooth regions image to be detected
Further include before the step of carrying out odontopathy identification to odontopathy identification model：
Contrast carried out to the tooth regions image to be detected and brightness evaluation obtains contrast and deviates angle value and bright
Degree deviates angle value；
Deviate when angle value is more than second threshold pair when the contrast deviates angle value and is more than first threshold and/or the brightness
The tooth regions image to be detected carries out enhancing processing.
Another embodiment according to the present invention, the odontopathy identification model are based on MASK-RCNN convolutional neural networks
Classifier training obtains, including：
Input multiple tooth regions sample images to the MASK-RCNN convolutional neural networks grader obtain it is described multiple
The odontopathy coarse positioning and classification results of tooth regions sample image；
Based on the odontopathy coarse positioning and multiple described tooth regions sample images of classification results filtering；
Color carried out to the filtered tooth regions sample image and Texture Matching handles to obtain treated odontopathy
Sample image；
Odontopathy identification model is established according to the odontopathy sample image.
Another embodiment according to the present invention, the MASK-RCNN convolutional neural networks grader includes upper layer network
With lower layer's network, the upper layer network is used to extract the candidate odontopathy location boundary frame of the tooth regions sample image, described
Lower layer's network is used to extract feature from each odontopathy location boundary frame and carries out classification and boundary recurrence, and the upper layer network is
Faster R-CNN networks with mask branch, lower layer's network are ResNet networks.
Another embodiment according to the present invention, it is described to the filtered tooth regions sample image carry out color and
Texture Matching handles to obtain the step of treated odontopathy sample image and includes：
Color characteristic matching is carried out to the filtered tooth regions sample image using histogram method；
Textural characteristics matching is carried out to the filtered tooth regions sample image using Tamura algorithms；
Odontopathy position is carried out to the filtered tooth regions sample image and odontopathy type marks to obtain odontopathy sample
Image.
On the other hand, the present invention also provides a kind of the tooth disease diagnosing equipment, including：
Acquisition module, for obtaining tooth regions image to be detected；
Identification module carries out odontopathy knowledge for the tooth regions image to be detected to be input to odontopathy identification model
Not, and odontopathy recognition result is exported, the odontopathy recognition result includes odontopathy position and odontopathy type.
In another aspect, the present invention also provides a kind of intelligent image harvester, including the acquisition portion of upper end and lower end
Hand-held part, is connecting rod between the hand-held part and acquisition portion, and the outside in the acquisition portion is provided with waterproof cover, internal setting
Image capture module, the hand-held part include being set to positive first button and being set to the second button of side.
According to an embodiment of the present invention, described image acquisition module includes：Main processor unit, with the main process task
The image acquisition units of device module connection, gyro sensors unit, communication unit, Power Management Unit and are deposited at LED light filling units
Storage unit, the main processor unit includes arm processor and video encoding processor, for realizing Video coding processing, figure
As processing, Interface Controller and light filling control, described image collecting unit includes micro-lens and imaging sensor, the LED
Light filling unit includes blue LED lamp and white LED lamp, and the gyro sensors unit is used to measure to be turned in image acquisition process
Dynamic or defection signal, the communication unit are WiFi communication, and the Power Management Unit includes battery subelement and wireless charging
Subelement.
Another embodiment according to the present invention, the waterproof cover inside is waterproof inner shell, and the waterproof cover lower end is equipped with
Card slot, the card slot fix clamping, the external setting waterproof membrane and bulge loop of the waterproof cover and clamping end, the card with clamping end
Tight end is set to one end of elastic strip, and the other end and the connecting rod of the elastic strip are connected and fixed, the connecting rod with elasticity
One end of item connection is equipped with holding cylinder, is connected by stage body between the holding cylinder and connecting rod, the stage body and waterproof cover end
Face paste is tight, and hollow connection inner cylinder is additionally provided with inside the connecting rod；
The holding cylinder is equipped with the first straight slot and the second straight slot, is equipped with spacing ring in the holding cylinder, outer wall is equipped with
Sealing ring, the sealing ring are adjacent to sealing with waterproof cover inner wall, clamping cylinder, the clamping cylinder are equipped with inside the holding cylinder
Outside is equipped with the first card-tight part, the second card-tight part for being each passed through the first straight slot, the second straight slot and waterproof cover inner wall clamping, described
First straight slot, the second straight slot are separately positioned on sealing ring both sides up and down, and first straight slot and the second straight slot are uniformly arranged on guarantor
It holds in cylinder circumferential direction.
Beneficial effects of the present invention are as follows：
The tooth disease diagnosing method and device of the present invention, by obtaining tooth regions image to be detected, and it is input to tooth
Sick identification model carries out odontopathy and identifies to obtain odontopathy position and odontopathy type, the convenient diagnosis for realizing odontopathy of intelligence.The present invention
The tooth disease diagnosing method and device use the image dental diagnostic method based on deep learning, will need to pass through for a long time in conventional dental
The diagnosis method of discrimination for testing accumulation is converted into the automatic recognition classification method of machine learning, effectively improves the effect of auxiliary diagnosis
Rate realizes the automation of the tooth disease diagnosing.
Description of the drawings
Fig. 1 is a kind of flow diagram of one embodiment of the tooth disease diagnosing method of the present invention；
Fig. 2 is a kind of flow diagram of another embodiment of the tooth disease diagnosing method of the present invention；
Fig. 3 is a kind of flow signal of one embodiment of the odontopathy identification model training of the tooth disease diagnosing method of the present invention
Figure；
Fig. 4 is a kind of structure diagram of one embodiment of the tooth disease diagnosing equipment of the present invention；
Fig. 5 is a kind of structure diagram of another embodiment of the tooth disease diagnosing equipment of the present invention；
Fig. 6 is the structural schematic diagram of the MASK-RCNN convolutional neural networks of the present invention；
Fig. 7 is a kind of front view of the specific implementation mode of intelligent image harvester of the present invention；
Fig. 8 is a kind of left view of the specific implementation mode of intelligent image harvester of the present invention；
Fig. 9 is a kind of vertical view of the specific implementation mode of intelligent image harvester of the present invention；
Figure 10 is that a kind of waterproof cover of the specific implementation mode of intelligent image harvester of the present invention and link rod part dispense
Distribution structure schematic diagram；
Figure 11 is assembled portion A-A sectional views in Figure 10；
Figure 12 is clamping cylinder B-B sectional views in Figure 10 (after removal sealing ring)；
Figure 13 is a kind of structural schematic diagram of the image capture module of intelligent image harvester of the present invention.
Specific implementation mode
To keep the technical problem to be solved in the present invention, technical solution and advantage clearer, below in conjunction with attached drawing and tool
Body embodiment is described in detail.
On the one hand, as shown in Figs. 1-3, an embodiment of the present invention provides a kind of the tooth disease diagnosing methods, including：
Step 100：Obtain tooth regions image to be detected；
Step 200：Tooth regions image to be detected is input to odontopathy identification model and carries out odontopathy identification, and exports tooth
Sick recognition result, odontopathy recognition result include odontopathy position and odontopathy type.
The tooth disease diagnosing method of the embodiment of the present invention is by obtaining tooth regions image to be detected, and it is input to odontopathy
Identification model carries out odontopathy and identifies to obtain odontopathy position and odontopathy type, the convenient diagnosis for realizing odontopathy of intelligence.The present invention is real
The tooth disease diagnosing method for applying example uses the image dental diagnostic method based on deep learning, and protracted experience will be needed in conventional dental
The diagnosis method of discrimination of accumulation is converted into the automatic recognition classification method of machine learning, effectively improves the efficiency of auxiliary diagnosis,
Realize the automation of the tooth disease diagnosing.
As one for example, the step 100 of the tooth disease diagnosing method of the embodiment of the present invention includes：
Step 101：Image to be detected is acquired using intelligent image harvester；
Image to be detected of the embodiment of the present invention can be by special intelligent image collecting device, also can be by being peeped in oral cavity
The mobile terminals such as mirror, X-ray scanners or camera, mobile phone obtain to acquire.
It should be noted that the advantages of acquiring image using common camera is that adjustment mode is simple, it may make that image is complete
Office has more general visual effect, but can not carry out imaging control for specific dynamic object, especially in different illumination
Environmental condition under, tooth target can not be made to reach good imaging effect；Image is acquired using mobile terminals such as mobile phones
Advantage is that versatility is stronger, carries out auto-focusing for selection region, while carrying out brightness imaging control based on selection region,
But it is more demanding for the image quality of odontopathy as the auxiliary diagnosis of mobile terminal, thus for the clarity of image and
Luminance contrast has independent judging quota, meanwhile, the position size and operability of regional choice can all have final imaging
It influences, also has higher operation requirement to the shooting skill of user, be not suitable for being applicable in for generality；Stomatology endoscope, X-ray
The selection that scanner acquires image its photo environment, angle and pattern measurement point is carried out by manual measurement, image enhancement and
Processing is for the ease of DATA REASONING, and whole purpose is for doing data acquisition, and as to machine learning and automatic discrimination
Image and improper.Therefore, in order to ensure the quality of picture to be detected, it is preferred to use special intelligent image collecting device.
Step 102：Image to be detected is detected based on AdaBoost detection algorithms and whether there is tooth regions, if so, sentencing
It is tooth regions image to be detected to determine image to be detected.
In image acquisition process, it would be desirable to the position of monitoring tooth in real time, the tooth disease diagnosing side of the embodiment of the present invention
Method carries out target detection using AdaBoost algorithms.AdaBoost algorithms are that a series of Weak Classifier is combined into one strong point
Class device provides example set first, then carries out circulate operation to the example set, and cycle obtains a weak vacation first every time
If then calculating the error rate of the hypothesis, the weight for changing each example according to the error rate enters next cycle, algorithm
Flow is as follows：
1) the weights distribution of initialization training data.If there is N number of image pattern, then when each training sample most starts
All it is endowed identical weights：1/N；
2) training Weak Classifier.In specific training process, if some sample point is accurately classified, in structure
It makes in next training set, its weights are just lowered；On the contrary, if some sample point is not classified accurately, it
Weights be just improved.Then, the sample set that right value update is crossed be used to train next grader, and entire training process is such as
This is made iteratively down.
3) Weak Classifier that each training obtains is combined into strong classifier.The training process of each Weak Classifier terminates
Afterwards, the weight for increasing the small Weak Classifier of error in classification rate, makes it play larger decisive action in final classification function,
And the weight of the big Weak Classifier of error in classification rate is reduced, so that it is played smaller decisive action in final classification function.
The tooth disease diagnosing method of the embodiment of the present invention detects mapping to be checked by AdaBoost detection algorithms trained in advance
As whether there is tooth regions, basis is provided for the positioning of subsequent odontopathy position.
As another for example, further including before the step 200 of the tooth disease diagnosing method of the embodiment of the present invention：
Step 300：Contrast is carried out to tooth regions image to be detected and brightness evaluation obtains contrast and deviates angle value
Deviate angle value with brightness, deviates when angle value is more than second threshold pair when contrast deviates angle value and is more than first threshold and/or brightness
Tooth regions image to be detected carries out enhancing processing.
For odontopathy image, the image at odontopathy position has part and teeth space, gum, the root of the tongue from the point of view of color and shape
Or other normal positions taken are similar, simultaneously because Image Acquisition quality problems, part contrast is not very high
The odontopathy feature that odontopathy position is presented not is clearly, it is therefore desirable to do enhancing processing to image.The embodiment of the present invention
Retinex algorithm is enhanced using retina, various types of image is adaptively enhanced, it can be in dynamic model
It encloses, edge and three aspect of color reach balance.Retina enhancing Retinex algorithm is the color according to object, i.e., by object pair
What the radioactivity of light determined, and it is unrelated with intensity of illumination.Assuming that image S (x, y), is decomposed into R (x, y) and L (x, y), R
(x, y) is reflection subject image, and L (x, y) is incident light images, therefore, calculates the opposite relationship between light and dark between each pixel, can
It is corrected with each pixel to image, so that it is determined that the color of the pixel.
According to Retinex theories, human eye perceives illumination and body surface of the brightness of object depending on environment to irradiation light
Reflection, mathematic(al) representation is：
I (x, y)=L (x, y) * R (x, y) (1)
Both sides take logarithm：Log [R (x, y)]=log [I (x, y)]-log [L (x, y)] (2)
Wherein I (x, y) represents the picture signal that observed or camera receives；L (x, y) represents the sub-irradiation of ambient light
Amount；R (x, y) indicates to carry the reflecting component of the target object of image detail information.For image data I (x, y), calculate pair
The R (x, y) answered, then R (x, y) be considered enhanced image, L (x, y) can be by carrying out Gauss to image data I (x, y)
It is fuzzy to obtain.Algorithm detailed process is as follows：
(1) it calculates original image and carries out the image L (x, y) after obscuring by specified scale；
(2) Gaussian Blur that each scale is carried out to original image, the image Li (x, y) after being obscured, wherein small tenon i
Indicate scale parameter.
(3) to carrying out accumulation calculating under each scale
Log [R (x, y)]=Log [R (x, y)]+Weight (i) * (Log [Ii (x, y)]-Log [Li (x, y)]) (3)
Wherein Weight (i) indicates the corresponding weight of each scale, it is desirable that the sum of each scale weight is necessary for 1, classical
Value is equal weight.
(4) Log [R (x, y)] is quantified as the pixel value of 0 to 255 ranges, as final output.The mode of quantization is adopted
Equal interval quantizing is carried out then to each value Value with the maximum value Max and minimum M in for calculating Log [R (x, y)], it is public
Formula is：
R (x, y)=(Value-Min)/(Max-Min) * (255-0) (4)
Brightness is the index of most visual evaluation dental imaging imaging.Based on experience value, select the desired value of mean intensity for
100,10 irrelevances of the grade as object brightness of each decile between 0-100 and 100-255, setting weight is 0.4 i.e. the
One threshold value.For tooth, because of the influence of many factors such as the otherness of color, ambient lighting, simple brightness evaluation refers to
Number is insufficient as the completeness foundation of dental imaging reference, therefore introduces contrast evaluation simultaneously.Based on experience value, perfect condition
The desired value of lower contrast mean value is 25, therefore, by tooth contrast mean value from 0-25 etc. points of 10 grades, as target area
The irrelevance of contrast, weight are 0.15 i.e. second threshold.
The tooth disease diagnosing method of the embodiment of the present invention can be fed back evaluation result by the quality evaluation to image to be detected
Imaging control adjustment is carried out to camera, provides the shutter and gain imaging parameters of initialization deployment, later each frame upon start up
Imaging adjustment result can all carry out feedback and to camera carry out parameter setting, likewise, the image obtained every time can also be protected
The corresponding imaging parameters information of the frame image is stayed, the reference frame as follow-up imaging control.(1) when can't detect tooth at
As feature：First, target is not present in coverage, second is that since ambient lighting problem makes picture imaging excessive lightness or darkness, this
When can be used center survey light pattern, shooting camera imaging width be w, be highly h, the point centered on shooting center, obtain width
Degree is w/2, is highly the rectangle of h/2 as photometry region, calculates the brightness of image in the region, establishes buffer length as 10 frames
Regional luminance list, zoning luminance mean value.(2) tooth detects imaging features：With the luminance mean value of tooth sequence and right
It is used as dental imaging feature than degree mean value.Tooth brightness is calculated, the tooth brightness list that buffer length is 10 frames is established, calculates tooth
Tooth luminance mean value.Tooth histogram is counted, Threshold segmentation is done with maximum variance between clusters, calculates separately high portion's luminance mean value
Grayhigh and lower curtate luminance mean value graylow, with formula:
NContrast_LP=(grayhigh-graylow) * 100/256 (5)
Tooth contrast nContrast_LP is calculated, tooth contrast list is established, calculates tooth contrast list mean value.
(3) imaging parameters adjustment is after determining imaging control adjustable strategies, and multi-parameter collaboration adjusts gain and shutter so that mesh
Mark the imaging effect being optimal.Including brightness imaging control and contrast imaging control：
Brightness imaging control：When the brightness value for participating in calculating is less than brightness lower limit, if fast gate value is not transferred to maximum,
The luminance difference it is expected brightness and participate in calculating is calculated, luminance difference and shutter adjustment ratio mapping table (table 1) are searched, by adjustment
Ratio improves fast gate value, until maximum shutter；It is 1 increase gain according to adjustment amplitude, directly if shutter has been adjusted to the limit
To maximum gain；When participate in calculate brightness value be higher than the brightness upper limit when, if yield value be not transferred to it is minimum, according to adjustment walk
Long 1 reduces gain, until gain floor；If gain has been adjusted to lower limit, the luminance difference it is expected brightness and participate in calculating is calculated,
It searches luminance difference and shutter adjusts ratio mapping table, fast gate value is reduced in adjustment ratio, until fast gate value lower limit.
Table one：Luminance difference and shutter adjust ratio mapping table
Contrast imaging control module：When the contrast value for participating in calculating is less than contrast lower limit, if fast gate value does not have
It is transferred to maximum, calculates the contrast difference it is expected contrast and participate in calculating, contrast difference is searched and shutter adjusts ratio
Mapping table (table 2) improves fast gate value, until maximum shutter in adjustment ratio；If shutter has been adjusted to the limit, according to adjustment
Amplitude is 1 increase gain, until maximum gain；When the brightness value for participating in calculating is higher than the brightness upper limit, if yield value does not have
It is transferred to minimum, reduces gain according to adjusting step 1, until gain floor；If gain has been adjusted to lower limit, calculates and it is expected contrast
With the contrast difference for participating in calculating, searches contrast difference and shutter adjusts ratio mapping table, shutter is reduced in adjustment ratio
Value, until fast gate value lower limit；When carrying out contrast imaging control adjustment with reference target, as contrast is less than low threshold
Value, brightness are higher than high threshold, dim；Contrast is less than Low threshold, and brightness is less than Low threshold, lightens；It is compared when with tooth
When degree imaging control adjustment, if contrast is less than Low threshold, when brightness is higher than desired value, imaging does not adjust；Contrast
It is lightened when brightness is less than desired value less than Low threshold.
Table 2：Contrast difference and shutter adjust ratio mapping table
The tooth disease diagnosing method of the embodiment of the present invention carries out contrast evaluation and brightness to tooth regions image to be detected
Evaluation obtains contrast and deviates angle value and brightness deviation angle value, and when contrast deviates, angle value is more than first threshold and/or brightness is inclined
Enhancing processing is carried out to tooth regions image to be detected when being more than second threshold from angle value, to ensure the matter of dental imaging
Amount.
As another for example, the odontopathy identification model of the tooth disease diagnosing method of the embodiment of the present invention be based on
MASK-RCNN convolutional neural networks classifier trainings obtain, including：
Step 201：It inputs multiple tooth regions sample images to MASK-RCNN convolutional neural networks graders and obtains multiple
The odontopathy coarse positioning and classification results of tooth regions sample image；
Step 202：Multiple tooth regions sample images are filtered based on odontopathy coarse positioning and classification results；
Step 203：Color and Texture Matching are carried out to filtered tooth regions sample image and handle to obtain that treated
Odontopathy sample image；
Step 204：Odontopathy identification model is established according to odontopathy sample image.
The odontopathy identification model of the embodiment of the present invention is obtained based on MASK-RCNN convolutional neural networks classifier trainings, first
It first passes through MASK-RCNN convolutional neural networks grader and odontopathy coarse positioning and classification is carried out to multiple tooth regions sample images,
Screening is filtered to image by artificial or other methods later, face then is carried out to filtered tooth regions sample image
Color and Texture Matching handle to obtain treated odontopathy sample image, and odontopathy identification model is established based on odontopathy sample image.
The tooth disease diagnosing method of the embodiment of the present invention trains odontopathy identification model, model robust by the method for deep learning
Property it is strong, classification quick and precisely.
As another for example, the MASK-RCNN convolutional neural networks of the tooth disease diagnosing method of the embodiment of the present invention
Grader includes upper layer network and lower layer's network, and upper layer network is used to extract the candidate odontopathy position side of tooth regions sample image
Boundary's frame, lower layer's network are used to extract feature from each odontopathy location boundary frame and carry out classification and boundary recurrence, upper layer network
For the Faster R-CNN networks with mask branch, lower layer's network is ResNet networks.
As shown in fig. 6, in order to realize the quickly Accurate Segmentation in accurate odontopathy target detection and odontopathy region, based on built
Vertical odontopathy sample database, the embodiment of the present invention have selected the essence of MASK-RCNN convolutional neural networks graders progress odontopathy target
Really detection and segmentation.Faster R-CNN are that each candidate target output class label and frame offset, MaskR-CNN exist
The branch of a prediction segmentation mask on each area-of-interest is added on the basis of Faster R-CNN.
Faster R-CNN are made of two stages, and the first stage proposes that candidate target bounding box, second stage use
RoIPool extracts feature from each candidate frame and carries out classification and boundary recurrence.Mask R-CNN are in Faster R-CNN
On two-stage, parallel with prediction class and frame offset is that each RoI exports binary mask.On RoI after each sampling
Multitask loss function is defined as：L=Lcls+Lbox+Lmask
Wherein Lcls is Classification Loss, and Lbox loses for detection block, and Lmask is that average binary system intersects entropy loss.It is right
The definition of Lmask allows network to be that each class independently predicts binary mask, by special classification branch prediction for selecting
The class label of output masking.It is the two of m*m that, which there are each RoI in mask branches the output of Km2 dimensions, i.e. K classification, resolution ratio,
Value mask.Therefore use single pixel sigmoid two-value cross entropies, two-value cross entropy that the mask of every one kind can be made not competing mutually
It strives, rather than compares with the mask of other classes.
In Faster R-CNN, small characteristic pattern is extracted from each RoI using RoIPool, RoIPool is first by floating number
The RoI of expression is zoomed to finally summarizes each piece of covering with the matched granularity of characteristic pattern then by the RoI piecemeals after scaling
The characteristic value in region.Such calculating makes RoI and the feature of extraction misplace.Although this may not influence to classify, because of classification
There is certain robustness to transformation by a small margin, but it has prodigious negative effect to the accurate mask of prediction pixel grade.
Mask R-CNN propose RoIAlign layers to remove the dislocation of RoIPool, by the feature of extraction and input accurate alignment.
The exact value that bilinear interpolation calculates each position is used in RoIAlign, and result is summarized (using maximum or average pond
Change), to carry out the alignment of pixel to pixel pixel-to-pixel.
Classification and Detection for a variety of odontopathy illness images and segmentation, it is as follows that we construct MASK-RCNN networks：For with
In lower layer's convolutional network of the feature extraction in whole image, it is 50 layers of ResNet that we, which use depth, in MASK-RCNN
Feature is extracted from the final convolutional layer of fourth stage.For upper layer network, the Faster proposed in ResNet and FPN is extended
The upper layer network of R-CNN is added to a mask branch respectively.
As another for example, the step 203 of the tooth disease diagnosing method of the embodiment of the present invention includes：
Step 2031：Color characteristic matching is carried out to filtered tooth regions sample image using histogram method；
Step 2032：Textural characteristics matching is carried out to filtered tooth regions sample image using Tamura algorithms；
Step 2033：Odontopathy position is carried out to tooth regions sample image and odontopathy type marks to obtain odontopathy sample graph
Picture.
The method that the color-match of the embodiment of the present invention uses Histogram Matching, calculation formula are as follows：
H1And H2The respectively histogram of template image and overlapping region image calculates the Bhattacharyya of two histograms
Distance is as follows：
The Texture Matching of the embodiment of the present invention carries out textural characteristics matching using Tamura algorithms.Tamura textural characteristics
Six components correspond to six attribute of textural characteristics on Psychological Angle, are roughness (coarseness), contrast respectively
(contrast), direction degree (directionality), line picture degree (linelikeness), regularity (regularity) and thick
Slightly spend (roughness).The embodiment of the present invention mainly calculates odontopathy template texture and waits for using roughness, contrast and direction degree
The similarity of the target image of detection.
(1) roughness：
The calculating of roughness can be divided into following steps progress.First, it is 2k × 2k picture to calculate size in image
The average intensity value of pixel, that is, have in the active window of element
Wherein k=0,1 ..., 5 and g (i, j) is pixel intensity value positioned at (i, j).Then, for each pixel, respectively
It is poor to calculate its mean intensity in the horizontal and vertical directions between the window of non-overlapping copies.
Wherein for each pixel, the k values that E values reach maximum (no matter direction) can be made to be used for that optimum size Sbest is arranged
(x, y)=2k.Finally, roughness can be obtained by calculating the average value of Sbest in entire image, be expressed as：
(2) contrast
Contrast is obtained by the statistics to pixel intensity distribution situation, that is, passes through α4=μ4/σ4Come what is defined,
Middle μ4It is four squares and σ2It is variance.Contrast is weighed by following formula:
The value gives the global measurement of contrast in whole image or region.
(3) direction degree
The calculating of direction degree needs to calculate the gradient vector at each pixel first.The vector field homoemorphism and direction define respectively
For：
| Δ G |=(| ΔH|+|ΔV|)/2
θ=tan-1(ΔV/ΔH)+π/2 (12)
Wherein .H and .V be respectively as obtained by image convolution following two 3x3 operators both horizontally and vertically on
Variable quantity.
After the gradient vector of all pixels is all computed, a histogram HD is configured to expression θ values.This is straight
Side's figure carries out discretization to the codomain range of θ first, has then counted corresponding in each bin | .G | it is more than the picture of given threshold value
Prime number amount.This histogram can show peak value for the image with apparent directionality, then for the image without apparent direction
Show relatively flat.Finally, the directionality of image totality can be obtained by calculating the acuity of peak value in histogram, table
Show as follows:
P in above formula represents the peak value in histogram, npFor peak value all in histogram.For some peak value p, WpGeneration
All bin that the table peak value is included, andIt is the bin with peak.
On the other hand, as illustrated in figures 4-5, the embodiment of the present invention additionally provides a kind of the tooth disease diagnosing equipment, including：
Acquisition module 10, for obtaining tooth regions image to be detected；
Identification module 20 carries out odontopathy identification for tooth regions image to be detected to be input to odontopathy identification model,
And odontopathy recognition result is exported, odontopathy recognition result includes odontopathy position and odontopathy type.
As one for example, the acquisition module 10 of the tooth disease diagnosing equipment of the embodiment of the present invention includes：
Collecting unit 11, for utilizing intelligent image harvester to acquire image to be detected；
Detection unit 12 detects image to be detected with the presence or absence of tooth regions for being based on AdaBoost detection algorithms, if
It is then to judge that image to be detected is tooth regions image to be detected.
The tooth disease diagnosing equipment of the embodiment of the present invention obtains tooth regions image to be detected by acquisition module, and its is defeated
Enter to detection module, carrying out odontopathy by odontopathy identification model identifies to obtain odontopathy position and odontopathy type, and intelligence is convenient real
The diagnosis of existing odontopathy.The tooth disease diagnosing equipment of the embodiment of the present invention differentiates the diagnosis for needing protracted experience to accumulate in conventional dental
It is converted into the automatic recognition classification of machine learning, effectively improves the efficiency of auxiliary diagnosis, realizes the automation of the tooth disease diagnosing.
As another for example, the tooth disease diagnosing equipment of the embodiment of the present invention further includes：
Quality assessment module 30, for being compared to tooth regions image progress contrast to be detected and brightness evaluation
Degree deviates angle value and brightness and deviates angle value, and when contrast deviates, angle value is more than first threshold and/or brightness deviates angle value and is more than the
Enhancing processing is carried out to tooth regions image to be detected when two threshold values.
As another for example, the tooth disease diagnosing equipment of the embodiment of the present invention further includes odontopathy identification model training mould
Block 40, including：
Odontopathy coarse positioning and taxon 41, for inputting multiple tooth regions sample images to MASK-RCNN convolution god
The odontopathy coarse positioning and classification results of multiple tooth regions sample images are obtained through network classifier；
Filter element 42 filters multiple tooth regions sample images for being based on odontopathy coarse positioning and classification results；
Color and Texture Matching unit 43, for carrying out color and Texture Matching to filtered tooth regions sample image
Processing obtains that treated odontopathy sample image；
Model foundation unit 44, for establishing odontopathy identification model according to odontopathy sample image.
In another aspect, as illustrated in figures 7 to 13, the embodiment of the present invention additionally provides a kind of intelligent image harvester, including upper
The acquisition portion 210 at end and the hand-held part 110 of lower end are connecting rod 120, the external setting in acquisition portion between hand-held part and acquisition portion
There is waterproof cover 130, internal that image capture module 50 is arranged, hand-held part includes being set to positive first button 111 and setting
The second button 112 in side.
As one for example, the image capture module 50 of the embodiment of the present invention includes：Main processor unit 51, with master
The image acquisition units 52 of processor module connection, LED light filling units 53, gyro sensors unit 54, communication unit 55, power supply
Administrative unit 56 and storage unit 57, main processor unit 51 include arm processor and video encoding processor (not shown), are used
In realize Video coding processing, image procossing, Interface Controller and light filling control, image acquisition units 52 include micro-lens with
Imaging sensor (not shown), LED light filling units 53 include blue LED lamp and white LED lamp, and gyro sensors unit 54 is used for
The rotation in image acquisition process or defection signal are measured, communication unit 55 is WiFi communication, and Power Management Unit 56 includes electricity
Pond unit and wireless charging subelement (not shown).
As one for example, 130 inside of waterproof cover of the intelligent image harvester of the embodiment of the present invention is waterproof
Inner casing 131,130 lower end of waterproof cover are equipped with card slot 132, and card slot 132 fixes clamping, waterproof cover 130 and clamping end with clamping end 126
126 external setting waterproof membrane 140 and bulge loop 141, clamping end 126 are set to one end of elastic strip 125, elastic strip 125 it is another
One end is connected and fixed with connecting rod 120, and connecting rod 120 is equipped with holding cylinder 127, holding cylinder in the one end being connect with elastic strip 125
It is connected by stage body 122 between 127 and connecting rod 120, stage body 122 is adjacent to 130 end face of waterproof cover, and 120 inside of connecting rod is also
Equipped with hollow connection inner cylinder 121；
Holding cylinder 127 is equipped with the first straight slot 123 and the second straight slot 124, and spacing ring 150 is equipped in holding cylinder 127, outside
Wall is equipped with sealing ring 410, and sealing ring 410 is adjacent to sealing with 130 inner wall of waterproof cover, clamping cylinder is equipped with inside holding cylinder 127
310, clamping cylinder outside is equipped with the first card for being each passed through the first straight slot 123, the second straight slot 124 and 130 inner wall clamping of waterproof cover
Tight portion 311, the second card-tight part 312, the first straight slot 123, the second straight slot 124 are separately positioned on about 410 both sides of sealing ring, and first
Straight slot 123 and the second straight slot 124 are uniformly arranged in 127 circumferential direction of holding cylinder.
In use, limiting clamping wound packages by spacing ring enters position, then distinguished by first the second card-tight part of card-tight part
With the clamping of waterproof cover inner wall to ensure the assembly intensity of waterproof cover and connecting rod.
The conducting wire 220 of the image capture module of the embodiment of the present invention is connected to power supply with power supply circuit, using fold line (and electricity
It is similar to talk about line), it can elongate, stretch when use, conducting wire can be prevented to be torn in this way.
Be equipped with waterproof membrane outside the card slot of the embodiment of the present invention, waterproof membrane upper and lower ends respectively near card slot upper and lower ends
It is sealing assembly, fixed, and waterproof membrane is made of highly elastic material, such as silica gel, rubber, when needing to take out waterproof cover, it is only necessary to
Clamping end is pressed by waterproof membrane, directly pulls out waterproof cover after so that clamping end is released card slot.Waterproof membrane with clamping end
Corresponding position is equipped with bulge loop, and whens use can further press clamping end by bulge loop, is pressed to clamping end so that increasing
Distance is pressed, taking-up waterproof cover is facilitated.
The elastic strip of the embodiment of the present invention is made of elastic material, can be flexible plastic sheet, in use, passing through its bullet
Property makes clamping end be clamped in card slot.
The above is the preferred embodiment of the present invention, it is noted that for those skilled in the art
For, without departing from the principles of the present invention, it can also make several improvements and retouch, these improvements and modifications
It should be regarded as protection scope of the present invention.
Claims (10)
1. a kind of the tooth disease diagnosing method, which is characterized in that including：
Obtain tooth regions image to be detected；
The tooth regions image to be detected is input to odontopathy identification model and carries out odontopathy identification, and exports odontopathy identification knot
Fruit, the odontopathy recognition result include odontopathy position and odontopathy type.
2. a kind of the tooth disease diagnosing method according to claim 1, which is characterized in that described to obtain tooth regions to be detected
The step of image includes：
Image to be detected is acquired using intelligent image harvester；
Described image to be detected is detected based on AdaBoost detection algorithms and whether there is tooth regions, if so, being waited for described in judgement
Detection image is tooth regions image to be detected.
3. a kind of the tooth disease diagnosing method according to claim 1, which is characterized in that described by institute
It states before tooth regions image to be detected is input to the step of odontopathy identification model carries out odontopathy identification and further includes：
Contrast is carried out to the tooth regions image to be detected and brightness evaluation obtains contrast deviation angle value and brightness is inclined
From angle value, when the contrast deviates, angle value is more than first threshold and/or the brightness deviates when angle value is more than second threshold pair
The tooth regions image to be detected carries out enhancing processing.
4. a kind of the tooth disease diagnosing method according to claim 1, which is characterized in that the odontopathy identification model be based on
MASK-RCNN convolutional neural networks classifier trainings obtain, including：
It inputs multiple tooth regions sample images to the MASK-RCNN convolutional neural networks grader and obtains multiple described teeth
The odontopathy coarse positioning and classification results of area sample image；
Based on the odontopathy coarse positioning and multiple described tooth regions sample images of classification results filtering；
Color carried out to the filtered tooth regions sample image and Texture Matching handles to obtain treated odontopathy sample
Image；
Odontopathy identification model is established according to the odontopathy sample image.
5. according to a kind of the tooth disease diagnosing method described in claim 4, which is characterized in that the MASK-RCNN convolutional neural networks
Grader includes upper layer network and lower layer's network, and the upper layer network is used to extract the candidate tooth of the tooth regions sample image
Sick location boundary frame, lower layer's network are used to extract feature from each odontopathy location boundary frame and carry out classification and boundary time
Return, the upper layer network is the Faster R-CNN networks with mask branch, and lower layer's network is ResNet networks.
6. a kind of the tooth disease diagnosing method stated according to claim 4, which is characterized in that described to the filtered tooth regions
Sample image carries out the step of color and Texture Matching handle to obtain treated odontopathy sample image and includes：
Color characteristic matching is carried out to the filtered tooth regions sample image using histogram method；
Textural characteristics matching is carried out to the filtered tooth regions sample image using Tamura algorithms；
Odontopathy position is carried out to the filtered tooth regions sample image and odontopathy type marks to obtain odontopathy sample image.
7. a kind of the tooth disease diagnosing equipment, which is characterized in that including：
Acquisition module, for obtaining tooth regions image to be detected；
Identification module carries out odontopathy identification for the tooth regions image to be detected to be input to odontopathy identification model, and
Odontopathy recognition result is exported, the odontopathy recognition result includes odontopathy position and odontopathy type.
8. a kind of intelligent image harvester, which is characterized in that the hand-held part in acquisition portion and lower end including upper end, it is described hand-held
It is connecting rod between portion and acquisition portion, the outside in the acquisition portion is provided with waterproof cover, internal that image capture module, institute is arranged
It includes being set to positive first button and being set to the second button of side to state hand-held part.
9. a kind of intelligent image harvester according to claim 8, which is characterized in that described image acquisition module packet
It includes：Main processor unit, the image acquisition units being connect with the main processor modules, LED light filling units, gyro sensors list
Member, communication unit, Power Management Unit and storage unit, the main processor unit include at arm processor and Video coding
Device is managed, is controlled for realizing Video coding processing, image procossing, Interface Controller and light filling, described image collecting unit includes
Micro-lens and imaging sensor, the LED light filling units include blue LED lamp and white LED lamp, the gyro sensors list
Member is WiFi communication, the power management for measuring rotation or defection signal in image acquisition process, the communication unit
Unit includes battery subelement and wireless charging subelement.
10. a kind of intelligent image harvester according to claim 8, which is characterized in that be anti-inside the waterproof cover
Water inner casing, the waterproof cover lower end are equipped with card slot, and the card slot fixes clamping with clamping end, and the waterproof cover is outer with clamping end
Waterproof membrane and bulge loop is arranged in portion, and the clamping end is set to one end of elastic strip, and the other end and the connecting rod of the elastic strip connect
Fixation is connect, the connecting rod is equipped with holding cylinder in the one end being connect with elastic strip, passes through platform between the holding cylinder and connecting rod
Body connects, and the stage body is adjacent to waterproof cover end face, and hollow connection inner cylinder is additionally provided with inside the connecting rod；
The holding cylinder is equipped with the first straight slot and the second straight slot, and spacing ring is equipped in the holding cylinder, and outer wall is equipped with sealing
Circle, the sealing ring and waterproof cover inner wall are adjacent to sealing, are equipped with clamping cylinder inside the holding cylinder, outside the clamping cylinder
Equipped with the first card-tight part, the second card-tight part for being each passed through the first straight slot, the second straight slot and waterproof cover inner wall clamping, described first
Straight slot, the second straight slot are separately positioned on sealing ring both sides up and down, and first straight slot and the second straight slot are uniformly arranged on holding cylinder
In circumferential direction.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810403336.0A CN108389207A (en) | 2018-04-28 | 2018-04-28 | A kind of the tooth disease diagnosing method, diagnostic device and intelligent image harvester |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810403336.0A CN108389207A (en) | 2018-04-28 | 2018-04-28 | A kind of the tooth disease diagnosing method, diagnostic device and intelligent image harvester |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108389207A true CN108389207A (en) | 2018-08-10 |
Family
ID=63065257
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810403336.0A Pending CN108389207A (en) | 2018-04-28 | 2018-04-28 | A kind of the tooth disease diagnosing method, diagnostic device and intelligent image harvester |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108389207A (en) |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109118485A (en) * | 2018-08-13 | 2019-01-01 | 复旦大学 | Digestive endoscope image classification based on multitask neural network cancer detection system early |
| CN109146897A (en) * | 2018-08-22 | 2019-01-04 | 北京羽医甘蓝信息技术有限公司 | Oral cavity radiation image method of quality control and device |
| CN109242011A (en) * | 2018-08-27 | 2019-01-18 | 深圳开立生物医疗科技股份有限公司 | A kind of method and device identifying image difference |
| CN109242032A (en) * | 2018-09-21 | 2019-01-18 | 桂林电子科技大学 | A kind of object detection method based on deep learning |
| CN109816669A (en) * | 2019-01-30 | 2019-05-28 | 云南电网有限责任公司电力科学研究院 | A kind of improvement Mask R-CNN image instance dividing method identifying power equipments defect |
| CN109949319A (en) * | 2019-03-12 | 2019-06-28 | 北京羽医甘蓝信息技术有限公司 | The method and apparatus of whole scenery piece permanent teeth identification based on deep learning |
| CN110047081A (en) * | 2019-03-26 | 2019-07-23 | 清华大学 | Example dividing method, device, equipment and the medium of chest x-ray image |
| CN110490860A (en) * | 2019-08-21 | 2019-11-22 | 北京大恒普信医疗技术有限公司 | Diabetic retinopathy recognition methods, device and electronic equipment |
| CN110856667A (en) * | 2018-08-22 | 2020-03-03 | 珠海格力电器股份有限公司 | Tooth cleaning method, apparatus, device and storage medium |
| CN110872861A (en) * | 2018-08-31 | 2020-03-10 | 魔水科技（北京）有限公司 | Intelligent water spraying rod, intelligent toilet lid and flushing system integrated with visual function |
| CN110895702A (en) * | 2018-09-12 | 2020-03-20 | 深圳云天励飞技术有限公司 | Clothing attribute identification detection method and device |
| CN111161275A (en) * | 2018-11-08 | 2020-05-15 | 腾讯科技（深圳）有限公司 | Method and device for segmenting target object in medical image and electronic equipment |
| CN111563887A (en) * | 2020-04-30 | 2020-08-21 | 北京航空航天大学杭州创新研究院 | Intelligent analysis method and device for oral cavity image |
| CN111611815A (en) * | 2020-04-07 | 2020-09-01 | 魏丽 | Adjacent tooth target detection platform of implant |
| CN111611848A (en) * | 2020-04-02 | 2020-09-01 | 北京中科虹霸科技有限公司 | Cadaver iris identification method and device |
| CN111700698A (en) * | 2020-05-14 | 2020-09-25 | 先临三维科技股份有限公司 | Dental scanning method, apparatus, system and computer readable storage medium |
| CN112450861A (en) * | 2019-09-06 | 2021-03-09 | 广达电脑股份有限公司 | Tooth area identification system |
| CN112634246A (en) * | 2020-12-28 | 2021-04-09 | 深圳市人工智能与机器人研究院 | Oral cavity image identification method and related equipment |
| CN113194872A (en) * | 2018-12-17 | 2021-07-30 | 株式会社森田制作所 | Identification device, scanner system, identification method, and identification program |
| CN113728363A (en) * | 2019-02-06 | 2021-11-30 | 3 形状股份有限公司 | Method for generating dental model based on objective function |
| US11389131B2 (en) | 2018-06-27 | 2022-07-19 | Denti.Ai Technology Inc. | Systems and methods for processing of dental images |
| CN115191949A (en) * | 2022-07-28 | 2022-10-18 | 哈尔滨工业大学 | Dental disease diagnosis method and diagnosis instrument |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120148986A1 (en) * | 2010-12-13 | 2012-06-14 | Jiayong Yan | Method for identification of dental caries in polychromatic images |
| JP2013078395A (en) * | 2011-09-30 | 2013-05-02 | Advance Co Ltd | Dental image processing system |
| CN103957348A (en) * | 2014-05-20 | 2014-07-30 | 段益群 | Underwater network camera |
| CN204004960U (en) * | 2014-08-10 | 2014-12-10 | 宁波肯德姆工贸有限公司 | A kind of water pistol joint |
| CN104253982A (en) * | 2014-10-20 | 2014-12-31 | 国家电网公司 | Underground observation instrument for cable conduit in power well and application method |
| CN204829060U (en) * | 2015-08-25 | 2015-12-02 | 陈建峰 | Novel fast plug -in card lock joint head of latch closure formula |
| CN105151128A (en) * | 2015-09-01 | 2015-12-16 | 盐城大捷实业有限公司 | Automobile panel clamping structure |
| KR101839789B1 (en) * | 2017-08-01 | 2018-03-19 | 주식회사 뷰노 | System for generating interpretation data of dental image |
| CN107965630A (en) * | 2017-12-26 | 2018-04-27 | 宁波华斯特林电机制造有限公司 | A kind of quick coupling |
-
2018
- 2018-04-28 CN CN201810403336.0A patent/CN108389207A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120148986A1 (en) * | 2010-12-13 | 2012-06-14 | Jiayong Yan | Method for identification of dental caries in polychromatic images |
| JP2013078395A (en) * | 2011-09-30 | 2013-05-02 | Advance Co Ltd | Dental image processing system |
| CN103957348A (en) * | 2014-05-20 | 2014-07-30 | 段益群 | Underwater network camera |
| CN204004960U (en) * | 2014-08-10 | 2014-12-10 | 宁波肯德姆工贸有限公司 | A kind of water pistol joint |
| CN104253982A (en) * | 2014-10-20 | 2014-12-31 | 国家电网公司 | Underground observation instrument for cable conduit in power well and application method |
| CN204829060U (en) * | 2015-08-25 | 2015-12-02 | 陈建峰 | Novel fast plug -in card lock joint head of latch closure formula |
| CN105151128A (en) * | 2015-09-01 | 2015-12-16 | 盐城大捷实业有限公司 | Automobile panel clamping structure |
| KR101839789B1 (en) * | 2017-08-01 | 2018-03-19 | 주식회사 뷰노 | System for generating interpretation data of dental image |
| CN107965630A (en) * | 2017-12-26 | 2018-04-27 | 宁波华斯特林电机制造有限公司 | A kind of quick coupling |
Cited By (32)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11389131B2 (en) | 2018-06-27 | 2022-07-19 | Denti.Ai Technology Inc. | Systems and methods for processing of dental images |
| CN109118485A (en) * | 2018-08-13 | 2019-01-01 | 复旦大学 | Digestive endoscope image classification based on multitask neural network cancer detection system early |
| CN110856667A (en) * | 2018-08-22 | 2020-03-03 | 珠海格力电器股份有限公司 | Tooth cleaning method, apparatus, device and storage medium |
| CN109146897A (en) * | 2018-08-22 | 2019-01-04 | 北京羽医甘蓝信息技术有限公司 | Oral cavity radiation image method of quality control and device |
| CN109146897B (en) * | 2018-08-22 | 2021-08-03 | 北京羽医甘蓝信息技术有限公司 | Oral cavity radiation image quality control method and device |
| CN110856667B (en) * | 2018-08-22 | 2021-03-26 | 珠海格力电器股份有限公司 | Tooth cleaning device, apparatus and storage medium |
| CN109242011A (en) * | 2018-08-27 | 2019-01-18 | 深圳开立生物医疗科技股份有限公司 | A kind of method and device identifying image difference |
| CN110872861A (en) * | 2018-08-31 | 2020-03-10 | 魔水科技（北京）有限公司 | Intelligent water spraying rod, intelligent toilet lid and flushing system integrated with visual function |
| CN110895702B (en) * | 2018-09-12 | 2021-01-12 | 深圳云天励飞技术有限公司 | Clothing attribute identification detection method and device |
| CN110895702A (en) * | 2018-09-12 | 2020-03-20 | 深圳云天励飞技术有限公司 | Clothing attribute identification detection method and device |
| CN109242032B (en) * | 2018-09-21 | 2021-11-16 | 桂林电子科技大学 | Target detection method based on deep learning |
| CN109242032A (en) * | 2018-09-21 | 2019-01-18 | 桂林电子科技大学 | A kind of object detection method based on deep learning |
| CN111161275A (en) * | 2018-11-08 | 2020-05-15 | 腾讯科技（深圳）有限公司 | Method and device for segmenting target object in medical image and electronic equipment |
| CN111161275B (en) * | 2018-11-08 | 2022-12-23 | 腾讯科技（深圳）有限公司 | Method and device for segmenting target object in medical image and electronic equipment |
| CN113194872A (en) * | 2018-12-17 | 2021-07-30 | 株式会社森田制作所 | Identification device, scanner system, identification method, and identification program |
| CN109816669A (en) * | 2019-01-30 | 2019-05-28 | 云南电网有限责任公司电力科学研究院 | A kind of improvement Mask R-CNN image instance dividing method identifying power equipments defect |
| CN113728363A (en) * | 2019-02-06 | 2021-11-30 | 3 形状股份有限公司 | Method for generating dental model based on objective function |
| CN109949319A (en) * | 2019-03-12 | 2019-06-28 | 北京羽医甘蓝信息技术有限公司 | The method and apparatus of whole scenery piece permanent teeth identification based on deep learning |
| CN110047081A (en) * | 2019-03-26 | 2019-07-23 | 清华大学 | Example dividing method, device, equipment and the medium of chest x-ray image |
| CN110490860A (en) * | 2019-08-21 | 2019-11-22 | 北京大恒普信医疗技术有限公司 | Diabetic retinopathy recognition methods, device and electronic equipment |
| CN112450861A (en) * | 2019-09-06 | 2021-03-09 | 广达电脑股份有限公司 | Tooth area identification system |
| CN112450861B (en) * | 2019-09-06 | 2022-06-17 | 广达电脑股份有限公司 | Tooth area identification system |
| CN111611848A (en) * | 2020-04-02 | 2020-09-01 | 北京中科虹霸科技有限公司 | Cadaver iris identification method and device |
| CN111611848B (en) * | 2020-04-02 | 2024-02-06 | 北京中科虹霸科技有限公司 | Cadaver iris recognition method and device |
| CN111611815A (en) * | 2020-04-07 | 2020-09-01 | 魏丽 | Adjacent tooth target detection platform of implant |
| CN111563887B (en) * | 2020-04-30 | 2022-04-22 | 北京航空航天大学杭州创新研究院 | Intelligent analysis method and device for oral cavity image |
| CN111563887A (en) * | 2020-04-30 | 2020-08-21 | 北京航空航天大学杭州创新研究院 | Intelligent analysis method and device for oral cavity image |
| CN111700698B (en) * | 2020-05-14 | 2022-07-08 | 先临三维科技股份有限公司 | Dental scanning method, apparatus, system and computer readable storage medium |
| CN111700698A (en) * | 2020-05-14 | 2020-09-25 | 先临三维科技股份有限公司 | Dental scanning method, apparatus, system and computer readable storage medium |
| CN112634246A (en) * | 2020-12-28 | 2021-04-09 | 深圳市人工智能与机器人研究院 | Oral cavity image identification method and related equipment |
| CN112634246B (en) * | 2020-12-28 | 2023-09-12 | 深圳市人工智能与机器人研究院 | Oral cavity image recognition method and related equipment |
| CN115191949A (en) * | 2022-07-28 | 2022-10-18 | 哈尔滨工业大学 | Dental disease diagnosis method and diagnosis instrument |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108389207A (en) | A kind of the tooth disease diagnosing method, diagnostic device and intelligent image harvester | |
| CN208172859U (en) | A kind of intelligent image acquisition device | |
| CN108197546B (en) | Illumination processing method and device in face recognition, computer equipment and storage medium | |
| CN108875821A (en) | The training method and device of disaggregated model, mobile terminal, readable storage medium storing program for executing | |
| CN110168562B (en) | Depth-based control method, depth-based control device and electronic device | |
| CN106778695A (en) | A kind of many people's examing heartbeat fastly methods based on video | |
| CN110519489A (en) | Image-pickup method and device | |
| CN103764013A (en) | Automatic exposure control device, control device, endoscope device and automatic exposure control method | |
| CN107886484A (en) | U.S. face method, apparatus, computer-readable recording medium and electronic equipment | |
| JP2007534179A (en) | Adjusting the brightness of digital images | |
| CN107800965B (en) | Image processing method, device, computer readable storage medium and computer equipment | |
| CN105608700B (en) | Photo screening method and system | |
| CN107911625A (en) | Light measuring method, device, readable storage medium storing program for executing and computer equipment | |
| JP2012508423A (en) | Video infrared retinal image scanner | |
| CN109635871A (en) | A kind of capsule endoscope image classification method based on multi-feature fusion | |
| CN111937082A (en) | Guidance method and system for remote dental imaging | |
| CN106859595A (en) | Tongue picture acquisition methods, device and system | |
| CN109089041A (en) | Recognition methods, device, electronic equipment and the storage medium of photographed scene | |
| CN105426843A (en) | Single-lens palm vein and palmprint image acquisition apparatus and image enhancement and segmentation method | |
| CN110363087B (en) | Long-baseline binocular face in-vivo detection method and system | |
| CN113824884B (en) | Shooting method and device, shooting equipment and computer readable storage medium | |
| CN109886170A (en) | A kind of identification of oncomelania intelligent measurement and statistical system | |
| CN109584210A (en) | Multispectral three-dimensional vein imaging system | |
| CN108093170B (en) | User photographing method, device and equipment | |
| CN109242792A (en) | A kind of white balance proofreading method based on white object |
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 |