CN109223178A - Hysteroscope intelligence edge calculations system with target positioning function - Google Patents
Hysteroscope intelligence edge calculations system with target positioning function Download PDFInfo
- Publication number
- CN109223178A CN109223178A CN201810996144.5A CN201810996144A CN109223178A CN 109223178 A CN109223178 A CN 109223178A CN 201810996144 A CN201810996144 A CN 201810996144A CN 109223178 A CN109223178 A CN 109223178A
- Authority
- CN
- China
- Prior art keywords
- video
- target
- light source
- key frame
- parameter
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B34/00—Computer-aided surgery; Manipulators or robots specially adapted for use in surgery
- A61B34/20—Surgical navigation systems; Devices for tracking or guiding surgical instruments, e.g. for frameless stereotaxis
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B34/00—Computer-aided surgery; Manipulators or robots specially adapted for use in surgery
- A61B34/20—Surgical navigation systems; Devices for tracking or guiding surgical instruments, e.g. for frameless stereotaxis
- A61B2034/2046—Tracking techniques
- A61B2034/2065—Tracking using image or pattern recognition
Landscapes
- Health & Medical Sciences (AREA)
- Surgery (AREA)
- Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Medical Informatics (AREA)
- Robotics (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Molecular Biology (AREA)
- Animal Behavior & Ethology (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Veterinary Medicine (AREA)
- Image Analysis (AREA)
- Endoscopes (AREA)
Abstract
The present invention provides a kind of hysteroscope intelligence edge calculations system with target positioning function, which includes integrated cavity mirror system and dynamic object positioning system, integrated cavity mirror system include multiple functional modules and central control unit；Multiple functional modules include pneumoperitoneum instrument, human-computer interaction screen, cold light source and video camera；Dynamic object positioning system is used to execute S100, obtains the video that the video camera is converted to；S200, key frame images are selected from video；S300, the YOLO target detection model that each key frame images input is preset to training, obtain multiple images identified with target posting and target category；S400, multiple images identified with target posting and target category are synthesized, obtains target positioning video, and the target positioning video is shown.The present invention can be effectively improved " noise " tender subject.
Description
Technical field
The present invention relates to hysteroscope technical fields, and in particular to a kind of hysteroscope intelligence edge calculations with target positioning function
System.
Background technique
Currently, surgical operation hysteroscopeization has become more and more popular, Minimally Invasive Surgery has become surgeon and patient
Common recognition.Cavity mirror system is capable of providing the operation screen of high definition amplification, the fine structure of in-vivo tissue can be clearly showed that, with tradition
Open surgery is compared, and the visual field is apparent, therefore it is more accurate, fine to perform the operation, effectively prevent other than operative site internal organs by
Unnecessary interference, and intraoperative hemorrhage is few, performs the operation safer.
During endoscope-assistant surgery, since hysteroscope shakes the noise characteristic in the bring Minimally Invasive Surgery visual field in mobile treatment,
Positioning for target and certain influence is generated to the excavation of target signature.
Summary of the invention
(1) the technical issues of solving
In view of the deficiencies of the prior art, the hysteroscope intelligence edge calculations with target positioning function that the present invention provides a kind of
System can be effectively improved " noise " tender subject.
(2) technical solution
In order to achieve the above object, the present invention is achieved by the following technical programs:
The present invention provides a kind of hysteroscope intelligence edge calculations system with target positioning function, and the computing system includes
Integrated cavity mirror system and dynamic object positioning system, in which:
The integration cavity mirror system includes the center of multiple functional modules and the multiple functional module work of control
Control unit；The multiple functional module includes pneumoperitoneum instrument, human-computer interaction screen, cold light source and video camera, the pneumoperitoneum instrument, described
Human-computer interaction screen, the cold light source and the video camera are connected to the central control unit, the cold light source and described take the photograph
Camera is all connected with endoscope optical, and the cold light source provides light source for the endoscope optical, and the video camera is by the optics
The collected optical signal of endoscope is converted to video, and the video is sent to the human-computer interaction screen progress video and is shown；
The dynamic object positioning system is used to execute: S100, obtaining the video that the video camera is converted to；S200,
According to the time of frame image each in the video and color, key frame images are selected from the video；S300, each is closed
The YOLO target detection model of training is preset in the input of key frame image, obtains what multiple were identified with target posting and target category
Image；S400, multiple described images identified with target posting and target category are synthesized, obtains target positioning view
Frequently, and by the target positioning video show；Wherein, the training process of the YOLO target detection model includes at least:
Training sample data collection is clustered using K-centers clustering method.
(3) beneficial effect
In the hysteroscope intelligence edge calculations system with target positioning function that the embodiment of the invention provides a kind of, dynamic mesh
Mark positioning system obtains hysteroscope video first, then extracts key frame images therein, then using YOLO trained in advance
Target detection model position to the target in key frame images and determines target type, then will have target posting and mesh
The image of mark classification logotype is synthesized, and dynamic target positioning video is obtained.Due to YOLO target detection mould trained in advance
It include being clustered using K-centers clustering method to training sample data collection, and use K- in the training process of type
The mode of centers cluster can be effectively improved " noise " tender subject, so as to improve the picture matter of target positioning video
Amount.For the present invention due to using target detection model to carry out the identification of target positioning and target type, treatment effeciency is high, locates simultaneously
It is fast to manage speed, can accomplish real-time target positioning and target type discrimination.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with
It obtains other drawings based on these drawings.
Fig. 1 shows the structure of the hysteroscope intelligence edge calculations system in one embodiment of the invention with target positioning function
Schematic diagram；
Fig. 2 shows the flow diagrams of method performed by dynamic object positioning system in one embodiment of the invention.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art
Every other embodiment obtained without creative efforts, shall fall within the protection scope of the present invention.
In a first aspect, the present invention provides a kind of hysteroscope intelligence edge calculations system with target positioning function, such as Fig. 1 institute
Show, the computing system includes integrated cavity mirror system and dynamic object positioning system, in which:
The integration cavity mirror system includes the center of multiple functional modules and the multiple functional module work of control
Control unit；The multiple functional module includes pneumoperitoneum instrument, human-computer interaction screen, cold light source and video camera, the pneumoperitoneum instrument, described
Human-computer interaction screen, the cold light source and the video camera are connected to the central control unit, the cold light source and described take the photograph
Camera is all connected with endoscope optical, and the cold light source provides light source for the endoscope optical, and the video camera is by the optics
The collected optical signal of endoscope is converted to video, and the video is sent to the human-computer interaction screen progress video and is shown；
As shown in Fig. 2, the dynamic object positioning system is used to execute: S100, obtaining what the video camera was converted to
Video；S200, the time according to frame image each in the video and color, select key frame images from the video；
S300, the YOLO target detection model that the input of each key frame images is preset to training, obtain multiple with target posting and
The image of target category mark；S400, multiple described images identified with target posting and target category are synthesized,
Target positioning video is obtained, and the target positioning video is shown；Wherein, the training of the YOLO target detection model
Process includes at least: being clustered using K-centers clustering method to training sample data collection.
It will be appreciated that above-mentioned central control unit is properly termed as CCU, 32 single-chip microcontrollers can be used, are applicable in from invasive
Communication criterion, central control unit as host, each functional module as slave, central control unit can by 485 interfaces,
485 hubs are communicated with pneumoperitoneum instrument, cold light source, video camera and human-computer interaction screen, central control unit specifically can by pneumoperitoneum instrument,
The running state information of the functional modules such as cold light source, endoscope is integrated and is packaged, and is sent to human-computer interaction screen, and be shown in
Relevant information (for example, video) can also be sent to the end PC by DVI communication interface by human-computer interaction screen, central control unit
(for example, Demonstration Classroom, meeting room).
Wherein, communication specification, that is, communication protocol, using RTU mode, data frame structure are as follows: frame head (1 byte), address (1
A byte), data length (1 byte), instruction (1 byte), data (N number of byte), CRC check (2 bytes) composition (refer to
Enable and all being sent with hexadecimal), it is specified that the various command formats of host, the response format of slave, the address of slave, host
Broadcasting format.
Each functional module and central control unit carry out the control of each functional module below as described below:
(1) the pneumoperitoneum instrument may include proportional control valve, switch electromagnetic valve, mass-air-flow sensor, baroceptor and
Gas bleeder valve, in which: the proportional control valve, the switch electromagnetic valve, the mass-air-flow sensor, the baroceptor and institute
It states gas bleeder valve to be arranged on supply air line, and is connect with the central control unit；The supply air line is the pneumoperitoneum instrument
To the pipeline of patient's body lumen input gas；The mass-air-flow sensor is used to detect the throughput parameter of the supply air line；Institute
Baroceptor is stated for detecting the pneumatic parameter of the supply air line；
Corresponding, the central control unit is for obtaining the throughput parameter and the pneumatic parameter, in the gas
When parameter being pressed to be less than preset first pneumatic parameter, exported according to the throughput parameter and preset Standard Gases flow parameter
Pwm signal, the pwm signal by switch time ratio of the metal-oxide-semiconductor to the control valve for being adjusted, to realize
Throughput parameter in the supply air line is adjusted；It is also used to be greater than or equal to first gas in the pneumatic parameter
When pressing parameter, the proportional control valve and the switch electromagnetic valve are closed；It is also used to be greater than or equal to the in the pneumatic parameter
When two pneumatic parameters, gas bleeder valve is opened；Wherein, second pneumatic parameter is greater than first pneumatic parameter.
It will be appreciated that supply air line is referred to as output pipe, i.e., out of pneumoperitoneum instrument output gas to patient's body lumen
Gas piping.
It will be appreciated that gas here can be carbon dioxide gas.
It will be appreciated that pneumatic parameter is pressure, throughput parameter is gas velocity.
It will be appreciated that proportional control valve, switch electromagnetic valve, gas bleeder valve, mass-air-flow sensor, baroceptor are respectively provided with
It is connect on supply air line, and with central control unit, but proportional control valve can be single by metal-oxide-semiconductor and center control
Member connection.
It will be appreciated that central control unit is controlled using PWM algorithm by metal-oxide-semiconductor comparative example control valve,
Specifically: the mark of throughput parameter (for example, air velocity) and setting that central control unit detects mass-air-flow sensor
Quasi- throughput parameter is compared, if throughput parameter is less than the Standard Gases flow parameter, increases accounting for for the pwm signal of output
Empty ratio, to increase the pulse width of output end air-flow, to increase output gas flow；And if throughput parameter is greater than the Standard Gases
Flow parameter, then reduce the duty ratio of the pwm signal of output, to reduce the pulse width of output end air-flow, to reduce defeated
Air-flow out.As it can be seen that being carried out by the air-flow that supply air line output end may be implemented in the PWM algorithm that central control unit uses micro-
It adjusts, to be maintained near preset Standard Gases flow parameter.
Here, corresponding pwm signal is exported to metal-oxide-semiconductor according to comparison result, opening for proportional control valve is controlled by metal-oxide-semiconductor
Logical and closing, and then the switch time ratio of proportional control valve is controlled, so that the gas in supply air line passes through ratio control
A series of equal high-frequency impulse air-flow of amplitudes is obtained after valve processed, i.e. realization " exhale-stop " formula gas supply.Within the regular hour, this
The tolerance of a little high-frequency impulse air-flows is equivalent to the tolerance of conventional switch mode one switch bout, so can be realized flow in this way
Accurate control and gas transmission it is steady.Here, the switch time ratio of comparative example control valve is adjusted, and may be implemented to pulse air
The width of stream is adjusted, and then changes the air-flow size of supply air line output end.
It will be appreciated that the air-flow size of supply air line output end is adjusted by using PWM algorithm realization, in turn
The pressure of the output gas of supply air line is adjusted in realization, right since PWM algorithm is accomplished that the fine tuning to flow
The adjusting of output gas pressure is also fine tuning.In practical applications, the pneumatic parameter detected in baroceptor is (for example, pressure
In the case where being less than preset first pneumatic parameter by force), the adjusting of gas pressure can be realized indirectly using above-mentioned PWM algorithm.
It is possible that the pneumatic parameter (for example, pressure) that baroceptor detects occur exceeds or reaches preset first air pressure ginseng
Several situations, such situation are possible as the failure of pneumoperitoneum instrument a part and lead to not carry out by above-mentioned PWM algorithm
The air pressure of output gas is adjusted in the mode of fine tuning, can close proportional control valve and switch electromagnetic valve at this time, that is, stop
The output of supply air line, until the pneumatic parameter that baroceptor detects is again less than the first pneumatic parameter.
But when appearance can not reduce supply air line air pressure by way of closing proportional control valve and switch electromagnetic valve
The case where when, in order to avoid leading to the generation of dangerous situation because crossing hyperbar, gas bleeder valve can also be set in pneumoperitoneum instrument, let out
Air valve can be set on supply air line, and connect with central control unit, when central control unit determines that baroceptor is examined
When the pneumatic parameter measured is greater than or equal to the second pneumatic parameter, then gas bleeder valve is opened, starts to lose heart, until in supply air line
Air pressure is under the first pneumatic parameter.
It will be appreciated that the second pneumatic parameter is greater than the first pneumatic parameter.For example, the first pneumatic parameter is a setting value,
Second pneumatic parameter is the setting value+10mmHg, but under normal circumstances, the second pneumatic parameter is no more than 30mmHg.Work as gas
When the pneumatic parameter that pressure sensor detects is greater than or equal to the first pneumatic parameter, proportional control valve and switching control pilot are closed.
When the pneumatic parameter that baroceptor detects is greater than or equal to the second pneumatic parameter, then proportional control valve and switch are being closed
On the basis of control valve, gas bleeder valve is opened.
Certainly, proportional control valve and switch electromagnetic valve can be simultaneously closed off when being not required to gas supply, prevents proportional control valve
Inside appearance machinery is stuck, is not closed completely and leaks, and realizes the double protection to supply air line.
In practical applications, the baroceptor of the input gas atmosphere for detecting carbon dioxide tank can also be set,
To monitor the gas flow in carbon dioxide tank, alarm is issued when it is lower than particular value, reminds medical staff.
Metal-oxide-semiconductor is passed through using PWM algorithm due to being provided with proportional control valve in pneumoperitoneum instrument, and by central control unit
Realize that the switch proportional time of comparative example control valve is adjusted, thus realize the fine tuning to throughput parameter in supply air line,
And then fine tuning of the realization to air pressure, i.e., the gas flow in supply air line is adjusted by no ladder, and then realize to pneumoperitoneum pressure
It precisely and smoothly adjusts, solves air pressure instantaneous variation brought by the moment of switch valve opening and closing, video shake temperature, improve
Image quality.
(2) cold light source includes: light source module group, constant current plate and radiator, in which:
The light source module group includes LED bulb, light collecting barrel and light guide bundles；
The constant current plate and the radiator are connected to the central control unit, and the central control unit is for adopting
The constant current plate is controlled with pid algorithm, deviation adjusting is carried out to the output electric current of the light source module group, and monitor the light in real time
The light-source temperature of source mould group, and drive the radiator to radiate when light-source temperature is increased to certain value.
It will be appreciated that the light that light collecting barrel and light guide bundles can generate LED bulb is handled, so that the light of output is more
Add and satisfies the use demand.
It will be appreciated that guaranteeing that constant current plate can be quick since central control unit is using pid algorithm control output electric current
Deviation is adjusted, maintain the stability of constant current output.
It will be appreciated that since central control unit monitors heat of light source in real time, when heat of light source is increased to certain value
Radiator can be driven to reduce light heat, the light in Minimally Invasive Surgery is avoided and be radiated near wound and cause interior for a long time
The case where portion's tissue burn, occurs.
(3) video camera includes CCD camera and camera shooting mainboard, in which:
The CCD camera is used to the collected optical signal of the endoscope optical institute being converted to electric signal, described to take the photograph
As mainboard is used to the electric signal being converted to video, and the video is sent to human-computer interaction screen and dynamic object positioning system
System.
It will be appreciated that video camera can also other than video is sent to human-computer interaction screen by central control unit
Video (i.e. the original video of no-fix target) is sent to dynamic object positioning system, so that dynamic object positioning system is to it
In target positioned, dynamic object positioning system is treated with targeting information and target category information
The video obtained synthesized by image is shown, for medical staff's reference.
(4) human-computer interaction screen is referred to as monitor, can also show the function such as endoscope optical, pneumoperitoneum instrument, cold light source
Can module working condition, in this case, the central control unit is also used to the pneumoperitoneum instrument, the cold light source and described
The running state information of endoscope optical is integrated and is transmitted to being shown on the human-computer interaction screen.
Certainly, human-computer interaction screen can also have the input interface of the manipulation information of pneumoperitoneum instrument, cold light source, for medical staff
Pneumoperitoneum instrument, cold light source manipulation information are inputted.
It will be appreciated that each functional module becomes one, integration is realized, such as be integrated in cabinet, formed just
In the minimally invasive hysteroscope equipment of the intelligence of carrying and there is hysteroscope video processing function.Since the minimally invasive hysteroscope equipment of intelligence uses integral type
Structure, occupied area is small, mobile easy to carry, good compatibility, the Minimally Invasive Surgery that can be used under various environment.
It will be appreciated that the hardware of above-mentioned dynamic object positioning system can specifically include processor, memory and display
Device is stored with computer program in memory, and above-mentioned target positioning may be implemented when computer program is executed by processor
Method, the target positioning video shown on display.
It will be appreciated that above-mentioned hysteroscope camera lens is the component in endoscope optical, in insertion body cavity and the dirty inner cavity of device
It is interior directly to observe and shoot, therefore shooting obtains video and is referred to as hysteroscope video.
It will be appreciated that cavity mirror system can be laparoscope system, thoracoscope system, joint cavity mirror system, may be used also certainly
To be other cavity mirror systems.Transit chamber lens head carries out video capture to inside cavity.
It will be appreciated that according to the time of image and color extraction key frame, actually using the time of image and face
The method of the variation of color extracts key frame.Time change can fully demonstrate the global information of image, and color characteristic is able to reflect
The localized variation information of image.
For example, dynamic object positioning system can extract key frame images by following steps:
S201, using the first frame image in video as a key frame images, and enable d=2；
It will be appreciated that d indicates the frame number of image in the video, for example, d=2 indicates the 2nd frame image in video
Frame number.
S202, calculating
Wherein, SiFor the i-th frame image in the video, si=s (ti, ci), tiFor the i-th frame image in the video institute
The time point at place, ciFor the color matrix of the i-th frame image.
For example, s'2=s2-s1, s'3=(s2-s1)+(s3-s2)。
S203, judge s'dWhether corresponding preset threshold is greater than, wherein s'dCorresponding preset threshold is m* β, and m is current
The totalframes of key frame images, β are constant:
If so, using the d frame image in the video as a key frame images, and enter step S204；
Otherwise, S204 is entered step.
It will be appreciated that s 'dFor measuring the otherness between the image based on time and color change feature, s 'dIt is bigger
It indicates that the otherness between image is bigger, the high image of similitude can be rejected in this way, retain the apparent image conduct of otherness
Key frame images.
Here, by s 'dWhether the d frame image compared to determine in video with preset threshold is key frame images.
S204, judge that d is less than the totalframes of the video:
If so, the numerical value of d is increased by 1, and return step S202；
Otherwise, terminate key frame images extraction process；
Here, by the way that d compared with totalframes, is only just terminated key frame images when d is equal to totalframes and extracted
Journey, to realize the traversal to frame image each in video.
It is, of course, also possible to extract the key frame in video using other modes, above step S201~S204 is only wherein
A kind of concrete mode.
In some embodiments, dynamic object positioning system is also used to each key frame images input presetting training
Before YOLO target detection model, carried out according to edge black surround of the visual field parameter of the hysteroscope camera lens to the key frame images
Smoothing processing is filtered denoising to the image after smoothing processing using high-pass filter, and using median filter to filtering
Image after denoising is filtered enhancing.
It will be appreciated that the edge black surround to key frame images is smoothed, the hysteroscope of available sharpness of border
Image.It is filtered again using high-pass filter and median filter, obtains removing the noise in key frame images and retain pass
High frequency section in key frame image.The step mainly realizes the optimization processing to key frame images, this is of the invention for realizing
For basic object it is not necessary to, therefore in certain embodiments can not include the step.
Wherein, the training process of the YOLO target detection model includes at least: using K-centers clustering method to instruction
Practice sample data set to be clustered.
The above-mentioned detailed process clustered using K-centers clustering method to training sample data collection includes: each
Particle after iteration is chosen from the sample point of cluster, and selection standard is to select object nearest from average value in cluster as cluster
Center can be effectively improved " noise " tender subject in this way.Training sample data can be concentrated using K-centers clustering method
Practical collection (i.e. ground truth box) clustered, to find the statistical law of ground truth box.With poly-
Class number k is the number of candidate frame (i.e. anchor boxes), with the high-dimensional dimension for candidate frame of the width of the frame of k cluster centre
Degree.
However, K-means clustering method used in the YOLO neural network of prior art Central Plains is very sensitive to " noise ",
So the image under mobile hysteroscope is caused to there are problems that " noise ".In contrast, the present invention can be effectively improved " noise " sensitivity
Problem improves image quality.
In addition to this, there are also other differences for YOLO target detection model provided by the invention: the YOLO target inspection
Surveying includes pond layer in the network structure of model, and the pond layer can successively sort n activation primitive value from small to large, will
N weighted value successively sorts from small to large, and n weighted value is multiplied with corresponding activation primitive value respectively, calculates n multiplication knot
The average value of fruit, and using the average value as final activation primitive value.
The pond layer used in the present invention can be referred to as sort-pooling, specifically swash according to cumulative sequence arrangement n
Function living: { a1, a2, a3...an}(a1< a2< a3< ...), rather than select it is maximum that.With n weight { w1, w2,
w3...wnMultiply it by obtain n value, the average value of this n value is taken, i.e.,In this way, neural network is still
It can learn to correspond to { w1, w2, w3...wn}={ 0,0,0 ... 1 } good, old maximum pond, and subsequent layer can be with
More information is obtained, gradient flows through all values in one layer when backpropagation.Sort-pooling can be realized faster and better
Ground convergence, the Optimized Iterative time retains more pictorial information, while also protruding important image information, so that target
Positioning and the more accurate and treatment effeciency of identification are higher.
However, pond layer in the prior art be max-pooling, refer to choose n activation primitive in it is maximum that,
Delete other activation primitives.So max-pooling there are loss of spatial information, the letter from multiple activation primitive cannot be used
The problems such as breath and backpropagation can only improve maximum pond activation primitive.
Here, by may be implemented to the excellent of target detection model to K-centers clustering method and sort-pooling
Change.
Dynamic object positioning system is determined using YOLO target detection model and is identified with target posting and target category
The process of image can specifically include:
S301, each key frame images are divided into S*S grid, s is the integer greater than 1；
S302, it is directed to each grid, determines that position, confidence level and the target category of target are general using multiple candidate frames
The corresponding confidence level of each candidate frame and the target category probability multiplication are obtained the candidate frame of the network by rate
In target belong to the confidence score of each target category；
S303, the corresponding candidate frame of confidence score that will be less than preset threshold filter out, and retain and are optionally greater than the default threshold
The corresponding candidate frame of confidence score of value；
S304, each candidate frame retained in each key frame images is carried out at non-maxima suppression (i.e. NMS)
Reason obtains the image identified with target posting and target category；Wherein, the target posting and the target class
It Biao Shi not correspond.
S400, multiple described images identified with target posting and target category are synthesized, it is fixed obtains target
Digital video.
In practical application, target category can be identified to the side that target posting is arranged in, with the classification to target
It is illustrated.
It is realized it will be appreciated that method performed by object locating system is based on image processing techniques, target therein
It can according to need setting, for example, the intracorporal some abnormal conditions of chamber, can identify exception by above-mentioned object locating system
The position of situation and the type of abnormal conditions.
As it can be seen that dynamic object positioning system obtains hysteroscope video first, key frame images therein are then extracted, then
The target in key frame images is carried out using YOLO target detection model trained in advance to position and determine target type, then will
The image identified with target posting and target category is synthesized, and dynamic target positioning video is obtained.Due to instructing in advance
It include being carried out using K-centers clustering method to training sample data collection in the training process of experienced YOLO target detection model
Cluster, and it can be effectively improved " noise " tender subject by the way of K-centers cluster, so as to improve target positioning
The image quality of video.Identification of the present invention due to carrying out target positioning and target type using target detection model simultaneously,
Treatment effeciency is high, processing speed is fast, can accomplish real-time target positioning and target type discrimination, significantly reduce clinical doctor
Raw workload, adjuvant clinical doctor perform the operation the real-time hard objectives position in art, and operating time is greatly shortened.
It should be noted that, in this document, relational terms such as first and second and the like are used merely to a reality
Body or operation are distinguished with another entity or operation, are deposited without necessarily requiring or implying between these entities or operation
In any actual relationship or order or sequence.Moreover, the terms "include", "comprise" or its any other variant are intended to
Non-exclusive inclusion, so that the process, method, article or equipment including a series of elements is not only wanted including those
Element, but also including other elements that are not explicitly listed, or further include for this process, method, article or equipment
Intrinsic element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that
There is also other identical elements in process, method, article or equipment including the element.
The above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Although with reference to the foregoing embodiments
Invention is explained in detail, those skilled in the art should understand that: it still can be to aforementioned each implementation
Technical solution documented by example is modified or equivalent replacement of some of the technical features；And these modification or
Replacement, the spirit and scope for technical solution of various embodiments of the present invention that it does not separate the essence of the corresponding technical solution.
Claims (10)
1. a kind of hysteroscope intelligence edge calculations system with target positioning function, which is characterized in that the computing system includes
Integrated cavity mirror system and dynamic object positioning system, in which:
The integration cavity mirror system includes the center control of multiple functional modules and the multiple functional module work of control
Unit；The multiple functional module includes pneumoperitoneum instrument, human-computer interaction screen, cold light source and video camera, the pneumoperitoneum instrument, described man-machine
Interaction screen, the cold light source and the video camera are connected to the central control unit, the cold light source and the video camera
It is all connected with endoscope optical, the cold light source provides light source for the endoscope optical, and the video camera will be peeped in the optics
The collected optical signal of mirror is converted to video, and the video is sent to the human-computer interaction screen progress video and is shown；
The dynamic object positioning system is used to execute: S100, obtaining the video that the video camera is converted to；S200, basis
The time of each frame image and color, select key frame images from the video in the video；S300, by each key frame
The YOLO target detection model of training is preset in image input, obtains multiple figures identified with target posting and target category
Picture；S400, multiple described images identified with target posting and target category are synthesized, obtains target positioning view
Frequently, and by the target positioning video show；Wherein, the training process of the YOLO target detection model includes at least:
Training sample data collection is clustered using K-centers clustering method.
2. computing system according to claim 1, which is characterized in that the pneumoperitoneum instrument includes proportional control valve, switch electricity
Magnet valve, mass-air-flow sensor, baroceptor and gas bleeder valve, in which: the proportional control valve, the switch electromagnetic valve, described
Mass-air-flow sensor, the baroceptor and the gas bleeder valve are arranged on supply air line, and single with the center control
Member connection；The supply air line is the pipeline that the pneumoperitoneum instrument inputs gas to patient's body lumen；The mass-air-flow sensor is used for
Detect the throughput parameter of the supply air line；The baroceptor is used to detect the pneumatic parameter of the supply air line；
Corresponding, the central control unit is joined for obtaining the throughput parameter and the pneumatic parameter in the air pressure
When number is less than preset first pneumatic parameter, according to the throughput parameter and preset Standard Gases flow parameter output PWM letter
Number, the pwm signal is for being adjusted the switch time ratio of the control valve, to realize in the supply air line
Throughput parameter is adjusted；It is also used to when the pneumatic parameter is greater than or equal to first pneumatic parameter, described in closing
Proportional control valve and the switch electromagnetic valve；It is also used to open when the pneumatic parameter is greater than or equal to the second pneumatic parameter
Gas bleeder valve；Wherein, second pneumatic parameter is greater than first pneumatic parameter.
3. computing system according to claim 1, which is characterized in that the cold light source include: light source module group, constant current plate and
Radiator, in which:
The light source module group includes LED bulb, light collecting barrel and light guide bundles；
The constant current plate and the radiator are connected to the central control unit, and the central control unit is for using
Pid algorithm controls the constant current plate and carries out deviation adjusting to the output electric current of the light source module group, and monitors the light source in real time
The light-source temperature of mould group, and drive the radiator to radiate when light-source temperature is increased to certain value.
4. computing system according to claim 1, which is characterized in that the video camera includes CCD camera and camera shooting master
Plate, in which:
The CCD camera is used to the collected optical signal of the endoscope optical institute being converted to electric signal, the camera shooting master
Plate is used to be converted to the electric signal video, and the video is sent to the human-computer interaction screen and target positioning system
System；
Corresponding, the human-computer interaction screen when receiving the video for being shown；The object locating system is used for
Dynamic object positioning is carried out when receiving the video.
5. computing system according to claim 1, which is characterized in that the central control unit is also used to logical by DVI
The video is sent to the end PC and is shown by letter interface.
6. computing system according to claim 1, which is characterized in that the central control unit is also used to the pneumoperitoneum
The running state information of instrument, the cold light source and the endoscope optical is integrated and is transmitted to the human-computer interaction screen
On shown；And/or the human-computer interaction screen is also used to show the defeated of the operation information of the pneumoperitoneum instrument and the cold light source
Enter interface.
7. described in any item computing systems according to claim 1~6, which is characterized in that the dynamic object positioning system from
The process that key frame images are selected in the video includes:
S201, using the first frame image in video as a key frame images, and enable d=2；
S202, calculating
S203, judge s'dWhether corresponding preset threshold is greater than, wherein s'dCorresponding preset threshold is m* β, and m is current key
The totalframes of frame image, β are constant:
If so, using the d frame image in the video as a key frame images, and enter step S204；
Otherwise, S204 is entered step；
S204, judge that d is less than the totalframes of the video:
If so, the numerical value of d is increased by 1, and return step S202；
Otherwise, terminate key frame images extraction process；
Wherein, SiFor the i-th frame image in the video, si=s (ti, ci), tiIt is locating in the video for the i-th frame image
Time point, ciFor the color matrix of the i-th frame image.
8. described in any item computing systems according to claim 1~6, which is characterized in that the dynamic object positioning system is also
For before the YOLO target detection model of training is preset in the input of each key frame images, according to the view of the hysteroscope camera lens
Wild parameter is smoothed the edge black surround of the key frame images, using high-pass filter to the image after smoothing processing
It is filtered denoising, and enhancing is filtered to the image after filtering and noise reduction using median filter.
9. described in any item computing systems according to claim 1~6, which is characterized in that the dynamic object positioning system will
The YOLO target detection model of training is preset in each key frame images input, obtains multiple with target posting and target category
The process of the image of mark includes:
S301, each key frame images are divided into S*S grid, s is the integer greater than 1；
S302, it is directed to each grid, position, confidence level and the target category probability of target is determined using multiple candidate frames, it will
The corresponding confidence level of each candidate frame and the target category probability multiplication, obtain the mesh in the candidate frame of the network
Mark belongs to the confidence score of each target category；
S303, the corresponding candidate frame of confidence score that will be less than preset threshold filter out, and retain optionally greater than the preset threshold
The corresponding candidate frame of confidence score；
S304, non-maxima suppression processing is carried out to each candidate frame retained in each key frame images, obtains one
The image identified with target posting and target category；Wherein, the target posting and the target category identify one by one
It is corresponding.
10. described in any item computing systems according to claim 1~6, which is characterized in that the YOLO target detection model
It include pond layer in network structure, the pond layer can successively sort n activation primitive value from small to large, by n weight
Value successively sorts from small to large, and n weighted value is multiplied with corresponding activation primitive value respectively, calculates the flat of n multiplied result
Mean value, and using the average value as final activation primitive value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810996144.5A CN109223178B (en) | 2018-08-29 | 2018-08-29 | Cavity mirror intelligent edge computing system with target positioning function |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810996144.5A CN109223178B (en) | 2018-08-29 | 2018-08-29 | Cavity mirror intelligent edge computing system with target positioning function |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109223178A true CN109223178A (en) | 2019-01-18 |
| CN109223178B CN109223178B (en) | 2021-04-06 |
Family
ID=65068898
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810996144.5A Active CN109223178B (en) | 2018-08-29 | 2018-08-29 | Cavity mirror intelligent edge computing system with target positioning function |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109223178B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109363736A (en) * | 2018-09-10 | 2019-02-22 | 安徽省胸科医院 | Configure the integrated cavity mirror system of central control module |
| CN114424974A (en) * | 2022-01-12 | 2022-05-03 | 四川大学华西医院 | Method and system for positioning laparoscope operation hole based on pneumoperitoneum simulation |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023060241A1 (en) * | 2021-10-08 | 2023-04-13 | Spiro Robotics, Inc. | Navigation and control for airway management procedures, assemblies and systems |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103985114A (en) * | 2014-03-21 | 2014-08-13 | 南京大学 | Surveillance video person foreground segmentation and classification method |
| CN104545774A (en) * | 2015-01-30 | 2015-04-29 | 合肥德铭电子有限公司 | Integrated, mobile and minimally invasive electronic endoscope system |
| CN107590456A (en) * | 2017-09-06 | 2018-01-16 | 张栖瀚 | Small micro- mesh object detection method in a kind of high-altitude video monitoring |
| CN107871119A (en) * | 2017-11-01 | 2018-04-03 | 西安电子科技大学 | A kind of object detection method learnt based on object space knowledge and two-stage forecasting |
| CN107943837A (en) * | 2017-10-27 | 2018-04-20 | 江苏理工学院 | A kind of video abstraction generating method of foreground target key frame |
-
2018
- 2018-08-29 CN CN201810996144.5A patent/CN109223178B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103985114A (en) * | 2014-03-21 | 2014-08-13 | 南京大学 | Surveillance video person foreground segmentation and classification method |
| CN104545774A (en) * | 2015-01-30 | 2015-04-29 | 合肥德铭电子有限公司 | Integrated, mobile and minimally invasive electronic endoscope system |
| CN107590456A (en) * | 2017-09-06 | 2018-01-16 | 张栖瀚 | Small micro- mesh object detection method in a kind of high-altitude video monitoring |
| CN107943837A (en) * | 2017-10-27 | 2018-04-20 | 江苏理工学院 | A kind of video abstraction generating method of foreground target key frame |
| CN107871119A (en) * | 2017-11-01 | 2018-04-03 | 西安电子科技大学 | A kind of object detection method learnt based on object space knowledge and two-stage forecasting |
Non-Patent Citations (6)
| Title |
|---|
| SAHIL SINGLA，黄小天，路雪: "新型池化层sort_pool2d实现更快更好的收敛：表现优于最大池化层（附代码实现）", 《HTTP://WWW.SOHU.COM/A/195961723_465975》 * |
| YALI ZHENG ET AL.: "Localisation of colorectal polyps by convolutional neural network features learnt from white light and narrow band endoscopic images of multiple databases", 《IEEE》 * |
| 丁振国: "基于K-center和信息增益的Web搜索结果聚类方法", 《计算机应用研究》 * |
| 周纪强: "监控视频中多类目标检测与多目标跟踪算法研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
| 陈响: "面向公交车场景的视频关键帧提取算法研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
| 高峰: "K-center聚类算法在教学评估中的应用", 《计算机工程与应用》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109363736A (en) * | 2018-09-10 | 2019-02-22 | 安徽省胸科医院 | Configure the integrated cavity mirror system of central control module |
| CN114424974A (en) * | 2022-01-12 | 2022-05-03 | 四川大学华西医院 | Method and system for positioning laparoscope operation hole based on pneumoperitoneum simulation |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109223178B (en) | 2021-04-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109171605A (en) | Intelligent edge calculations system with target positioning and hysteroscope video enhancing processing function | |
| WO2020224470A1 (en) | Medical endoscope image identification method and system, and endoscope image system | |
| US20210205030A1 (en) | Image capturing of the areas outside the abdomen to improve placement and control of a surgical device in use | |
| CN109223178A (en) | Hysteroscope intelligence edge calculations system with target positioning function | |
| CN109523522A (en) | Processing method, device, system and the storage medium of endoscopic images | |
| JP7412528B2 (en) | Modifying data from surgical robotic systems | |
| CN109124782A (en) | Intelligent integral cavity mirror system | |
| JP7166430B2 (en) | Medical image processing device, processor device, endoscope system, operating method and program for medical image processing device | |
| CN111863209B (en) | Colonoscopy quality assessment workstation based on image recognition | |
| CN109124702A (en) | Configure the intelligent cavity mirror system of pneumoperitoneum control and central control module | |
| CN109241898A (en) | Object localization method and system, the storage medium of hysteroscope video | |
| Okamoto et al. | Real-time identification of blood regions for hemostasis support in laparoscopic surgery | |
| US20230346392A1 (en) | Endoscopic image analysis and control component of an endoscopic system | |
| CN110742690A (en) | Method for configuring endoscope and terminal equipment | |
| CN102341032A (en) | System for use in surgical procedures | |
| CN109350239A (en) | Intelligent integral robot cavity mirror system with target positioning function | |
| CN108926319B (en) | Mobile intelligent integrated endoscope system with high-definition video processing module | |
| CN109243586A (en) | Configure the hysteroscope intelligence edge calculations system of central control module | |
| CN114663293A (en) | Image enhancement method and device, electronic equipment and endoscope system | |
| CN109363614A (en) | Intelligent integral robot cavity mirror system with high definition video enhancing processing function | |
| CN109151275A (en) | Intelligent edge calculations system with real-time hysteroscope video enhancing processing function | |
| Song et al. | A novel endoscope system for position detection and depth estimation of the ureter | |
| CN109363613A (en) | Inhibit the intelligent movable integration cavity mirror system of function with electromagnetism | |
| CN109363615A (en) | Inhibit the mobile integrated cavity mirror system of function with pneumoperitoneum control and electromagnetism | |
| CN109171606A (en) | Intelligent integral robot cavity mirror system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |