CN109124782A - Intelligent integral cavity mirror system - Google Patents

Intelligent integral cavity mirror system Download PDF

Info

Publication number
CN109124782A
CN109124782A CN201810997377.7A CN201810997377A CN109124782A CN 109124782 A CN109124782 A CN 109124782A CN 201810997377 A CN201810997377 A CN 201810997377A CN 109124782 A CN109124782 A CN 109124782A
Authority
CN
China
Prior art keywords
video
image
value
training sample
blurred
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
Application number
CN201810997377.7A
Other languages
Chinese (zh)
Other versions
CN109124782B (en
Inventor
丁帅
杨善林
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hefei University of Technology
Original Assignee
Hefei University of Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Hefei University of Technology filed Critical Hefei University of Technology
Priority to CN201810997377.7A priority Critical patent/CN109124782B/en
Publication of CN109124782A publication Critical patent/CN109124782A/en
Application granted granted Critical
Publication of CN109124782B publication Critical patent/CN109124782B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B90/00—Instruments, implements or accessories specially adapted for surgery or diagnosis and not covered by any of the groups A61B1/00 - A61B50/00, e.g. for luxation treatment or for protecting wound edges
    • A61B90/36—Image-producing devices or illumination devices not otherwise provided for
    • A61B90/361—Image-producing devices, e.g. surgical cameras
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B17/00—Surgical instruments, devices or methods, e.g. tourniquets
    • A61B17/00234—Surgical instruments, devices or methods, e.g. tourniquets for minimally invasive surgery
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B17/00—Surgical instruments, devices or methods, e.g. tourniquets
    • A61B17/02—Surgical instruments, devices or methods, e.g. tourniquets for holding wounds open; Tractors
    • A61B17/0281—Abdominal wall lifters
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B90/00—Instruments, implements or accessories specially adapted for surgery or diagnosis and not covered by any of the groups A61B1/00 - A61B50/00, e.g. for luxation treatment or for protecting wound edges
    • A61B90/36—Image-producing devices or illumination devices not otherwise provided for
    • A61B90/37—Surgical systems with images on a monitor during operation
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B90/00—Instruments, implements or accessories specially adapted for surgery or diagnosis and not covered by any of the groups A61B1/00 - A61B50/00, e.g. for luxation treatment or for protecting wound edges
    • A61B90/36—Image-producing devices or illumination devices not otherwise provided for
    • A61B90/361—Image-producing devices, e.g. surgical cameras
    • A61B2090/3612—Image-producing devices, e.g. surgical cameras with images taken automatically
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B2560/00—Constructional details of operational features of apparatus; Accessories for medical measuring apparatus
    • A61B2560/02—Operational features

Abstract

The present invention provides a kind of intelligent integral cavity mirror system, the system includes incorporating chamber lens device and processing unit, and incorporating chamber lens device includes multiple functional modules, controls the central control unit of multiple functional modules work and provide the power supply device of power supply for multiple functional modules；Multiple functional modules include pneumoperitoneum instrument, human-computer interaction screen, cold light source and the video camera for inflating to cavity, pneumoperitoneum instrument includes proportional control valve, switch electromagnetic valve, gas bleeder valve, mass-air-flow sensor and baroceptor, and power supply device includes Switching Power Supply and multiple electromagnetic interference suppression circuits；Processing unit is for carrying out dynamic object positioning and being handled video to obtain clear video.The present invention can solve the too fast caused video jitter problem of air pressure instantaneous variation brought by the moment of switch valve opening and closing, it avoids transmitting electromagnetic interference signal mutually between each functional module, clearly hysteroscope video is provided in real time for operation, is effectively improved " noise " tender subject.

Description

Intelligent integral cavity mirror system
Technical field
The present invention relates to cavity mirror system technical fields, and in particular to a kind of intelligent integral cavity mirror 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.
But current cavity mirror system has the disadvantage in that
(1) during endoscope-assistant surgery, since the noise in the hysteroscope shake bring Minimally Invasive Surgery visual field in mobile treatment is special Property, positioning for target and certain influence is generated to the excavation of target signature；
(2) the problem of being operated due to doctor, it is possible to lead to the significantly shaking of hysteroscope camera lens, and then some moulds occur Frame is pasted, operation generates certain influence；
(3) in the state that abdominal pressure reaches balance, shake can be all generated in the moment that switch valve is opened and is closed, therefore The visual effects in operation can be seriously affected；
(4) radiation of the endoscope all-in-one machine using each functional module inside shielding box shielding to space, to improve To the anti-interference ability of space radiation.But the electromagnetic interference signal that each functional module itself generates at work passes through work electricity Source line is transmitted to other function module, interferes with each other so as to cause between each functional module, to influence the image matter of endoscope Amount.
Summary of the invention
(1) the technical issues of solving
In view of the deficiencies of the prior art, the present invention provides a kind of intelligent integral cavity mirror systems.
(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 intelligent integral cavity mirror system, and the system comprises incorporating chamber lens devices and processing dress It sets, in which:
The incorporating chamber lens device includes the center control of multiple functional modules, the multiple functional module work of control Unit and the power supply device of power supply is provided for the multiple functional module；The multiple functional module includes for inflating to cavity Pneumoperitoneum instrument, human-computer interaction screen, cold light source and video camera, the pneumoperitoneum instrument, the human-computer interaction screen, the cold light source and described Video camera is connected to the central control unit；The cold light source and the video camera are all connected with endoscope optical；It is described cold Light source provides light source for the endoscope optical；The collected optical signal of the endoscope optical is converted to view by the video camera Frequently, and the video is sent to the human-computer interaction screen and carries out the displaying of hysteroscope video；Wherein:
The pneumoperitoneum instrument includes proportional control valve, switch electromagnetic valve, gas bleeder valve, mass-air-flow sensor and baroceptor, Wherein: the proportional control valve, the switch electromagnetic valve, the gas bleeder valve, the mass-air-flow sensor and the air pressure sensing Device is arranged on supply air line, and is connect with the central control unit；The mass-air-flow sensor is described for detecting The throughput parameter of supply air line；The baroceptor is used to detect the pneumatic parameter of the supply air line；Wherein, the confession Air pipe is the pipeline that the pneumoperitoneum instrument inputs gas to patient's body lumen；The central control unit is for obtaining the throughput Parameter and the pneumatic parameter, when the pneumatic parameter is less than preset first pneumatic parameter, according to the throughput parameter With preset Standard Gases flow parameter output pwm signal, the pwm signal be used for the switch time ratio of the control valve into Row is adjusted, and is adjusted with realizing to the throughput parameter in the supply air line；Be also used to be greater than in the pneumatic parameter or When equal to first pneumatic parameter, the proportional control valve and the switch electromagnetic valve are closed；It is also used to join in the air pressure When number is greater than or equal to the second pneumatic parameter, gas bleeder valve is opened；Wherein, second pneumatic parameter is joined greater than first air pressure Number；
The power supply device includes Switching Power Supply and multiple electromagnetic interference suppression circuits；The Switching Power Supply includes successively connecting The first rectification circuit, transformer and multiple second rectification circuits connect；First rectification circuit is used for the exchange being inputted Unidirectional voltage needed for electricity is changed into the transformer；The transformer includes an armature winding and multiple secondary windings, institute It states armature winding to connect with the output end of first rectification circuit, the multiple secondary windings and the multiple second rectified current The input terminal on road connects one to one；Each second rectification circuit is used to for the output voltage for corresponding to secondary windings being changed into pair DC voltage needed for answering functional module；The multiple electromagnetic interference suppression circuit, input terminal and the multiple second rectification The output end of circuit connects one to one, and output end is used to connect one to one with the input terminal of the multiple functional module, For inhibiting the bi-directional electromagnetic between the multiple functional module and the Switching Power Supply to interfere；Wherein, the secondary windings, institute The quantity for stating the second rectification circuit, the electromagnetic interference suppression circuit and the functional module is all the same；
The processing unit is used to execute: S ' 100, obtaining the video that the video camera is converted to；S ' 200, institute is determined It states the corresponding cavity position of each frame image in video and according to readability classifies to obtain by each frame image described Clear image and blurred picture corresponding to the intracorporal different parts of chamber；S ' 300, network pair is fought using generation trained in advance The corresponding blurred picture in each position carries out deblurring and handles to obtain corresponding de-blurred image；S ' 400, by each frame clear image It is synthesized to obtain the clear video of the cavity sequentially in time with de-blurred image, and the clear video is shown Show；
The processing unit is also used to execute: S100, according in the video or the clear video each frame image when Between and color, select key frame images from the video or the clear video；S200, each key frame images are inputted The YOLO target detection model for presetting training obtains multiple images identified with target posting and target category；S300, general Multiple described images identified with target posting and target category are synthesized, and obtain target positioning video, and will be described Target positioning video is shown；Wherein, the training process of the YOLO target detection model includes at least: using K- Centers clustering method clusters training sample data collection.
(3) beneficial effect
The embodiment of the invention provides a kind of intelligent integral cavity mirror systems, have the effect that
(1) due to being provided with proportional control valve in pneumoperitoneum instrument, and by central control unit using PWM algorithm realization pair The switch proportional time of proportional control valve is adjusted, to realize the fine tuning to throughput parameter in supply air line, Jin Ershi Now to the fine tuning of air pressure, i.e., by no ladder adjust supply air line in gas flow, and then realize to pneumoperitoneum pressure precisely and It smoothly adjusts, solves the too fast caused video shake of air pressure instantaneous variation brought by the moment of switch valve opening and closing and ask Topic improves image quality；
(2) in Switching Power Supply, for each functional module, a secondary windings and second rectified current are provided with Road, and then form a power circuit for the functional module.For N number of functional module, it is mutually independent to form N item Power circuit, making cannot be by shared power supply line transmission interference signal between each functional module.On this basis, at each It is also provided with an electromagnetic interference suppression circuit in functional module and corresponding second rectification circuit, for inhibiting functional module certainly The electromagnetic interference that body generates is transmitted by working power line to Switching Power Supply, while also Switching Power Supply being inhibited to pass through working power line To functional module electromagnetic interference, that is to say, that inhibit electromagnetic interference two-way between functional module and Switching Power Supply, and then avoid Electromagnetic interference signal is transmitted mutually between each functional module；
(3) processing unit carries out mould to the blurred picture in the collected video of cavity mirror system by generating confrontation network Then the clear image obtained after original clear image and deblurring processing is synthesized to obtain sequentially in time by paste processing Clear video can solve the problem of image blurring as caused by equipment shake under adverse circumstances, effectively to perform the operation or checking Hysteroscope video, the accuracy of raising target positioning mitigate the diagnosis of doctor or burden of performing the operation in real time and clearly for journey offer；
(4) processing unit extracts the key frame images in above-mentioned video or clear video first, then using instruction in advance Experienced YOLO target detection model position to the target in key frame images and determines target type, then will determine with target Position frame and the image of target category mark are synthesized, and dynamic target positioning video is obtained.Due to YOLO mesh trained in advance Mark includes being clustered using K-centers clustering method to training sample data collection, and adopt in the training process of detection model It can be effectively improved " noise " tender subject with the mode that K-centers is clustered, so as to improve the picture of target positioning video Face quality.Identification of the present invention due to carrying out target positioning and target type using target detection model simultaneously, treatment effeciency Height, processing speed are fast, 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 structural schematic diagram of intelligent integral cavity mirror system in one embodiment of the invention；
Fig. 2 shows the structural schematic diagrams of switch unit in one embodiment of the invention；
Fig. 3 shows the structural schematic diagram of electromagnetic interference suppression circuit in one embodiment of the invention；
Fig. 4 shows the flow diagram of Fuzzy Processing in one embodiment of the invention；
Fig. 5 shows the training process schematic diagram that confrontation network is generated 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.
The present invention provides a kind of intelligent integral cavity mirror system, as shown in Figure 1, the system include incorporating chamber lens device and Processing unit, in which:
The incorporating chamber lens device includes the center control of multiple functional modules, the multiple functional module work of control Unit and the power supply device of power supply is provided for the multiple functional module；The multiple functional module includes for inflating to cavity Pneumoperitoneum instrument, human-computer interaction screen, cold light source and video camera, the pneumoperitoneum instrument, the human-computer interaction screen, the cold light source and described Video camera is connected to the central control unit；The cold light source and the video camera are all connected with endoscope optical；It is described cold Light source provides light source for the endoscope optical；The collected optical signal of the endoscope optical is converted to view by the video camera Frequently, and the video is sent to the human-computer interaction screen and carries out the displaying of hysteroscope video.
Processing unit is for carrying out target positioning and carrying out image procossing to obtain clear video.
Each section in system is described in detail below:
(1) it is directed to the control of each functional module and central control unit to each functional module
(1.1) it is directed to central control unit
Above-mentioned central control unit is properly termed as CCU, can use 32 single-chip microcontrollers, is applicable in from invasive communication criterion, in Entreat control unit to be used as host, each functional module as slave, central control unit can by 485 interfaces, 485 hubs and Pneumoperitoneum instrument, cold light source, video camera and the communication of human-computer interaction screen, central control unit specifically can by pneumoperitoneum instrument, cold light source, interior peep The running state information of the functional modules such as mirror is integrated and is packaged, and human-computer interaction screen is sent to, and is shown in human-computer interaction screen, Central control unit can also by DVI communication interface by relevant information (for example, video) be sent to the end PC (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.
Central control unit can specifically be handled each task using multi-task coordination method, specifically: center Control unit main task type to be treated includes: extraneous control information (for example, the control information of cold light source, pneumoperitoneum instrument Control information etc.) and warning information, internal control information (for example, automatic adjustment information of pneumoperitoneum instrument and feedback etc.), letter Cease upload task and display task.In order to more reasonably handle each task well, our use priority dispatching algorithms will be above-mentioned Task classification, according to priority determine first handling for task, priority is from high to low are as follows: the external world control information and warning information, Itself internal control information, information upload task and display task；It again include some small tasks in the task of each of the above type, Therefore we take round-robin algorithm to handle it in the task of same level-one.
(1.2) it is directed to above-mentioned pneumoperitoneum instrument
The pneumoperitoneum instrument includes proportional control valve, switch electromagnetic valve, gas bleeder valve, mass-air-flow sensor and baroceptor, Wherein:
The proportional control valve, the switch electromagnetic valve, the gas bleeder valve, the mass-air-flow sensor and the air pressure transmission Sensor is arranged on supply air line, and is connect with the central control unit；The mass-air-flow sensor is for detecting institute State the throughput parameter of supply air line；The baroceptor is used to detect the pneumatic parameter of the supply air line；Wherein, described Supply air line is the pipeline that the pneumoperitoneum instrument inputs gas to patient's body lumen；
The central control unit is small in the pneumatic parameter for obtaining the throughput parameter and the pneumatic parameter When preset first pneumatic parameter, according to the throughput parameter and preset Standard Gases flow parameter output pwm signal, institute Pwm signal is stated for the switch time ratio of the control valve to be adjusted, to realize to the air-flow in the supply air line Amount parameter is adjusted；It is also used to close the ratio when the pneumatic parameter is greater than or equal to first pneumatic parameter Control valve and the switch electromagnetic valve；It is also used to open and lose heart when the pneumatic parameter is greater than or equal to the second pneumatic parameter Valve；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 and the gas bleeder valve being subsequently noted are arranged at supply air line On, it is connect with central control unit, and can be connect by metal-oxide-semiconductor with central control unit.
It will be appreciated that central control unit is controlled using PWM algorithm comparative example control valve, specifically: in The standard throughput ginseng of throughput parameter (for example, air velocity) and setting that centre control unit detects mass-air-flow sensor Number is compared, if throughput parameter is less than the Standard Gases flow parameter, increases the duty ratio of the pwm signal of output, thus The pulse width for increasing output end air-flow, to increase output gas flow；And if throughput parameter be 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 output gas flow.It can See, is finely adjusted by the air-flow that supply air line output end may be implemented in the PWM algorithm that central control unit uses, to keep 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 gas in this way The accurate control and gas transmission of flow are steady.Here, the switch time ratio of comparative example control valve is adjusted, and may be implemented to arteries and veins The width of qi of chong channel ascending adversely 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, since PWM algorithm is accomplished that the fine tuning to throughput, Adjusting to output gas pressure is also fine tuning.In practical applications, detect in baroceptor pneumatic parameter (for example, Pressure) less than in the case where preset first pneumatic parameter, it can realize the tune of gas pressure indirectly using above-mentioned PWM algorithm Section.It is possible that the pneumatic parameter (for example, pressure) that baroceptor detects occur exceeds or reaches preset first gas The case where pressing parameter, such situation are possible as the failure of pneumoperitoneum instrument a part and lead to not through above-mentioned PWM algorithm The air pressure of output gas is adjusted in the mode being finely adjusted, and can close proportional control valve and switch electromagnetic valve at this time, i.e., 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.When the pneumatic parameter that baroceptor detects When more than or equal to the first pneumatic parameter, proportional control valve and switching control pilot are closed.When the air pressure that baroceptor detects When parameter is greater than or equal to the second pneumatic parameter, then on the basis of closing proportional control valve and switching control pilot, opens and lose heart Valve.
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.
Comparison is realized using PWM algorithm due to being provided with proportional control valve in pneumoperitoneum instrument, and by central control unit The switch proportional time of example control valve is adjusted, to realize the fine tuning to flow parameter in supply air line, and then realization pair The fine tuning of air pressure adjusts the gas flow in supply air line by no ladder, and then realize precisely and steady to pneumoperitoneum pressure Adjusting, solve switch valve opening and closing moment brought by air pressure instantaneous variation it is too fast caused by video jitter problem, mention High picture quality.
In some embodiments, incorporating chamber lens device can also include storage unit, in which:
The storage unit connects the central control unit, for storing the corresponding data set of multiple age brackets；
It is corresponding, the multiple age bracket options of human-computer interaction screen display, and for detecting the multiple age When any option in section option is selected, trigger signal is sent to the central control unit；
The central control unit is also used to when receiving the trigger signal, according to the option selected corresponding year Data set corresponding to age section, determines the air pressure recommendation and throughput recommendation for being directed to the age bracket, and the air pressure is pushed away It recommends value and the throughput recommendation is shown on the human-computer interaction screen.
In practical applications, all age group may include: newborn, children, adult, the elderly etc..
It in practical applications, may include a plurality of record content in the data set of each age bracket, every record content can be with Including air pressure, throughput, the access times under the gentle flow set of the air pressure, abdominal cavity Pressure Inflation to specified air pressure threshold value model Average time used and this record within the scope of gas mean temperature, abdominal cavity Pressure Inflation to specified air pressure threshold value when enclosing interior The timestamp of content.For example, { p:8, s:20, ts:20, c:20, t:480, st:1528969758 }.Wherein, p is pressure, and s is Throughput, ts are access times, and c is gas mean temperature, and t is the average time used, and st is timestamp.
It will be appreciated that medical staff can select any of multiple age bracket options to select on human-computer interaction screen , central control unit is extracted from storage unit by the corresponding data set of selection option, and then calculates throughput recommendation With air pressure recommendation, and two recommendations are shown on human-computer interaction screen, for medical staff's reference, for example, according to air pressure Above-mentioned first pneumatic parameter and/or the second pneumatic parameter is arranged in recommendation, and above-mentioned standard air-flow is arranged according to throughput recommendation Measure parameter.
Here, for each age bracket, corresponding throughput recommendation and air pressure recommendation can be all calculated, thus When calculating recommendation, it is contemplated that this feature of the difference of different age group, so that recommendation more has reference value.
Gas temperature on average in above-mentioned data set is depending on the temperature that repeated detection arrives, it is therefore desirable to setting temperature Sensor is spent, temperature sensor can be set on the supply air line, connect with the central control unit, for detecting State the gas temperature in supply air line.
Wherein, central control unit can determine that the air pressure recommendation and the throughput are recommended using following steps Value:
A1, the suitable score value that every record content is calculated using the first formula, first formula include:
In formula, g is to be suitble to score value, a1、a2And a3For weighted value, tsmaxFor access times in a plurality of record content Maximum value, tmaxFor time used average in a plurality of record content maximum value, c is to record abdominal cavity in content in this Gas mean temperature when within the scope of Pressure Inflation to specified air pressure threshold value, t are to record abdominal cavity Pressure Inflation in content in this The average time used within the scope of to specified air pressure threshold value, ts is to record the access times in content in this.
Wherein, the central control unit can specifically calculate the weighted value in first formula according to third formula a1, the third formula includes:
a1(t)=a0e-25(t-1)
In formula, t is all numbers in the case where being chronomere with week where the timestamp of current record content, timestamp Closer, all numbers at place are smaller.
For example, if the timestamp of current record content within nearest one week, t 1；If the time of current record content Stamp is within nearest two weeks, then t is 2.
Wherein, three weighted value a1、a2And a3And be 1, i.e. a1+a2+a3=1.
A2, using the air pressure in record content corresponding to the maximum value in obtained multiple suitable score values as the gas Press recommendation and using the throughput in the corresponding record content of the maximum value as throughput recommendation.
It will be appreciated that the more proof air pressures of access times and the combination of throughput are better in record content, it is average Temperature is better closer to body temperature, and the shorter inflate the time used the better, integrates, g value is higher, it was demonstrated that corresponding air pressure and The combination of throughput is more suitable for patient, thus here using maximum g value it is corresponding record content in air pressure as Air pressure recommendation, using the throughput in the corresponding record content of maximum g value as throughput recommendation.
In some embodiments, central control unit can also carry out the record content in the data set in storage unit It updates or deletes, to guarantee the data validity of data set.
Wherein, the process that central control unit is updated the data set in storage unit includes:
The pneumoperitoneum instrument in use, first time abdominal cavity air pressure and default pressure difference in specified threshold When maintaining preset duration, to current gas pressure, current throughput, current gas temperatures, this abdominal cavity Pressure Inflation to specified air pressure Time used and timestamp in threshold range are recorded；If in current record content including the current gas pressure and described The record of current throughput, then add 1 to corresponding access times, and be updated to timestamp, is calculated more using the second formula Average time used and gas mean temperature, second formula after new includes:
In formula, n is updated average time used or updated gas mean temperature, and o is being averaged before updating Gas mean temperature before time used or update, ts are access times before this update, and r is this abdominal cavity Pressure Inflation Time used or the current gas temperatures within the scope of to specified air pressure threshold value.
For example, the gas mean temperature at first 10 times is 37.2 DEG C, this gas temperature is 37.1 DEG C, then new gas Mean temperature is n=(37.2*9+37.1)/10.
Certainly, if not having current record content in storage unit, current record content is added in data set, And access times are 1.
Here, the update to access times is realized to the mode of access times plus 1, when replacing old using new timestamp Between the mode stabbed realize the update to timestamp, it is real by way of the old average time used is replaced the new average time used Now to the update of time averagely used, old gas mean temperature is replaced by new gas mean temperature, realization pair The update of gas mean temperature.
Wherein, the process that central control unit deletes the record content in data set in storage unit can wrap It includes: if the item number of the corresponding data centralized recording content of each age bracket of storage unit storage is greater than preset maximum item Number, then by the least note of access times in the record content of the smallest default number of branches of timestamp in the corresponding data set of the age bracket Content is recorded to delete.
That is, after the item number of the record content in the corresponding data set of some age bracket is greater than certain item number, Just the smallest record content of access times in the smallest a certain number of record contents of timestamp is deleted, it in this way can be by the time The smallest record content of earliest and access times is deleted.
In the conventional technology, in pneumoperitoneum instrument use process, medical staff needs to combine experience to pressure and throughput It is configured, is difficult in view of temperature and inflationtime the two factors, and here according to patient's gas replenishment process in historical data In gas temperature, inflationtime, pressure, throughput and access times group be combined into medical staff to provide suitable pressure gentle The combination of flow.
Comparative example is realized using PWM algorithm due to being provided with proportional control valve in pneumoperitoneum instrument, and by central control unit The switch proportional time of control valve is adjusted, to realize the fine tuning to throughput parameter in supply air line, and then realization pair The fine tuning of air pressure adjusts the gas flow in supply air line by no ladder, and then realize precisely and steady to pneumoperitoneum pressure Adjusting, solve switch valve opening and closing moment brought by air pressure instantaneous variation it is too fast caused by video shake temperature, mention High picture quality.
(1.3) it is directed to cold light source
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 light-source temperature in real time, when light-source temperature 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.
(1.4) it is directed to video camera
The 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 processing unit.
(1.5) it is directed to human-computer interaction screen
Human-computer interaction screen, is referred to as monitor, can also show the functions such as endoscope optical, pneumoperitoneum instrument, cold light source The working condition of module, in this case, the central control unit is also used to the pneumoperitoneum instrument, the cold light source and the light The running state information for learning endoscope 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.
(2) it is directed to power supply device
Power supply device includes Switching Power Supply and more electromagnetic interference suppression circuits；Wherein:
As shown in Fig. 2, the Switching Power Supply includes sequentially connected first rectification circuit 1, transformer 2 and multiple second whole Current circuit 5；Unidirectional voltage needed for the alternating current that first rectification circuit 1 is used to be inputted is changed into the transformer 2； The transformer 2 includes an armature winding 21 and multiple secondary windings 22, the armature winding 21 and first rectified current The output end on road 1 connects, and the input terminal of the multiple secondary windings 22 and the multiple second rectification circuit 5, which corresponds, to be connected It connects；Each second rectification circuit 5 is for the output voltage for corresponding to secondary windings 22 to be changed into needed for corresponding function module DC voltage；
The output end of the multiple electromagnetic interference suppression circuit, input terminal and the multiple second rectification circuit 5 is one by one It is correspondingly connected with, output end is used to connect one to one with the input terminal of multiple functional modules, for inhibiting the multiple function Bi-directional electromagnetic interference between module and the Switching Power Supply.
It will be appreciated that the alternating current of input can be converted to unidirectional high electricity needed for transformer 2 by the first rectification circuit 1 Pressure, for example, power supply grid is provided~AC220V converts.First rectification circuit 1 can use EMI and rectification circuit, That is the first rectification circuit 1 includes emi-filter module and rectification filtering module.Certainly, the first rectification circuit 1 may be used also To realize the conversion to alternating current using other circuit modules.
In practice, the output voltage of secondary windings 22 can be non-dc electricity, that is to say, that secondary windings 22 exported Voltage changes in certain waveform, but each voltage value on waveform is positive value.And each second rectification circuit 5 then will The voltage that corresponding secondary windings 22 exports is changed into direct current, so that corresponding functional module is used.
It will be appreciated that when power supply device provided by the invention is applied to endoscope all-in-one machine, in endoscope all-in-one machine There is N number of functional module, then N number of secondary windings 22 can be set in power supply device, N number of second rectification circuit 5 and N number of electricity are set Magnetic disturbance suppression circuit, N number of secondary windings 22 and N number of second rectification circuit 5 connect one to one, N number of second rectification circuit, 5 He N number of electromagnetic interference suppression circuit connects one to one.
It will be appreciated that Fig. 2 shows Switching Power Supply in, right side be Switching Power Supply input terminal, left side be switch electricity The output end in source.
In Switching Power Supply, for each functional module, it is provided with a secondary windings 22 and second rectified current Road 5, and then form a power circuit for the functional module.For N number of functional module, it is mutually indepedent to form N item Power circuit, making between each functional module cannot be by shared power supply line transmission interference signal.On this basis, each It is also provided with an electromagnetic interference suppression circuit in a functional module and corresponding second rectification circuit 5, for inhibiting function mould The electromagnetic interference that block itself generates is transmitted by working power line to Switching Power Supply, while Switching Power Supply also being inhibited to pass through work electricity Source line is to functional module electromagnetic interference, that is to say, that inhibits electromagnetic interference two-way between functional module and Switching Power Supply, in turn It avoids transmitting electromagnetic interference signal mutually between each functional module.
In some embodiments, Switching Power Supply in addition to include the first rectification circuit 1, transformer 2 and the second rectification circuit 5 it It outside, can also include secondary feedback unit 4 and the setting being arranged in the multiple secondary windings 22 on a secondary windings 22 Switch voltage-stabilizing control unit 3 on the armature winding 21, in which: the secondary feedback unit 4 be used for by it is described it is secondary around The voltage signal of group 22 feeds back to the switch voltage-stabilizing control unit 3, and the switch voltage-stabilizing control unit 3 is used for according to the electricity Signal is pressed to carry out pressure stabilizing control to the output voltage of secondary windings 22.
Here, voltage signal on secondary windings 22 is fed back to switch voltage-stabilizing control unit 3 by secondary feedback unit 4, in turn Switch voltage-stabilizing control unit 3 is adjusted according to output voltage of the voltage signal to secondary windings 22, to realize pressure stabilizing control, into The working performance of one step guarantee Switching Power Supply.
In some embodiments, electromagnetic interference suppression circuit can be realized by multiple structural forms, as shown in figure 3, wherein A kind of optional structure type are as follows: each electromagnetic interference suppression circuit includes sequentially connected Schottky diode 10, differential mode LC filter unit and common mode LC filter unit；Wherein:
The anode of the Schottky diode 10 connects the positive output end of corresponding second rectification circuit 5；
The differential mode LC filter unit includes differential mode inductance 11 and at least two first capacitors in parallel；The differential mode inductance 11 one end connects the cathode of the Schottky diode 10, and described at least two first capacitors in parallel are arranged in the differential mode Between the other end of inductance 11 and the negative output terminal of the second rectification circuit of the correspondence 5, second rectification circuit of correspondence 5 is born Output end ground connection；
The common mode LC filter unit includes common mode inductance 15 and at least two second capacitors in parallel, the common mode inductance One end of a winding connects the other end of the differential mode inductance 11 in 15, one end of another winding in the common mode inductance 15 The negative output terminal of second rectification circuit of correspondence 5 is connected, described at least two second capacitors in parallel are arranged in the common mode Between the other end of another winding of the other end and common mode inductance 15 of one winding of inductance 15；The common mode inductance Positive output end of the other end of 15 winding as the electromagnetic interference suppression circuit, the positive output end is for connecting corresponding function The positive input terminal of energy module；The other end of another winding of the common mode inductance 15 is as the negative of the electromagnetic interference suppression circuit Output end, the negative output terminal are used to connect the negative input end of corresponding function module.
It will be appreciated that Schottky diode 10 has the recovery of ultrafast electric current and unidirectional cutoff conductive characteristic, can only allow Switching Power Supply transmits signal to functional module, reduces and functional module is even prevented to transmit electricity to Switching Power Supply by working power line Magnetic disturbance signal can inhibit serviced functional module to pass through working power line electromagnetism caused by other function module in this way Interference.
In practical application, the differential mode inductance 11 in differential mode LC filter unit can use 10uH ring inductance, this electricity Feeling has the characteristic for preventing current break.The quantity of at least two first capacitors in parallel in differential mode LC filter unit can root According to needing to be arranged, for example, three first capacitors in parallel of setting, these three first capacitors can be using the NPO material electricity of 10uF The NPO material capacitor 14 for holding 12, the NPO material capacitor 13 of 100nF and 100pF, in this way can be in 1MHz to 300MHz frequency model There is lower equivalent resistance in enclosing, have good absorption inhibiting effect, such differential mode LC filter unit to electromagnetic interference signal Can effectively inhibit the external differential mode electromagnetic interference signal of 1MHz to 300MHz to imported into operating circuit, at the same to itself Good inhibiting effect is played in conduction of the electromagnetic interference signal to Switching Power Supply.
It will be appreciated that the common mode inductance 15 in common mode LC filter unit can believe the common mode coupled on working power line Number there is good inhibiting effect, external electromagnetic interference signal can be reduced and be coupled on the working power line of functional module, thus Inhibition is transmitted in the form of common mode interference by working power line.At least two second electricity in parallel in common mode LC filter unit The quantity of appearance can according to need selection, for example, two the second capacitors 16 and 17 in parallel of setting.Second capacitor can be used and be moved back Coupling high frequency capacitance.Common mode inductance 15 and the second capacitor 16,17 form filter circuit.
It will be appreciated that left side is the input of electromagnetic interference suppression circuit in the electromagnetic interference suppression circuit shown in Fig. 3 End, right side are the output end of electromagnetic interference suppression circuit.
Due on working power line DM EMI signal and common mode interference coexist, by Schottky diode 10, Differential mode LC filter unit, common mode LC filter unit combine the bi-directional electromagnetic codan to be formed, and both can effectively inhibit out Electromagnetic interference signal is transmitted to functional module by working power line in powered-down source, and functional module itself can effectively be inhibited to generate Electromagnetic interference transmitted by working power line to Switching Power Supply, and reduce the electromagnetic interference of functional module and pass through space radiation direction Outside transmission, improves the quality of the working power of functional module.
In the above-mentioned specific structure of electromagnetic interference suppression circuit, the negative output terminal of the second rectification circuit 5 is grounded, specific real Existing mode, which can be, is directly grounded the negative output terminal of the second rectification circuit 5, can also be the negative output of the second rectification circuit 5 Hold it is indirect earthed, for example, by the metal-back of the negative output terminal of the second rectification circuit 5 connection endoscope all-in-one machine, the metal-back It is connect by the protection ground wire of three-phase power line with power supply grid ground wire, to realize the negative output terminal ground connection of the second rectification circuit 5.
(3) it is directed to processing unit
(3.1) processing unit is also used to execute following methods, to realize the purpose for obtaining clear video:
S ' 100, the video that the video camera is converted to is obtained；
S ' 200, determine in the video the corresponding cavity position of each frame image and by each frame image according to Readability classifies to obtain clear image and blurred picture corresponding to the intracorporal different parts of the chamber；
It will be appreciated that including two kinds of classification in step S ' 200, one is divide cavity position belonging to image Class, one is divide the image into clear image and blurred picture.Both no sequencings of classifying, for example, can be first to figure As affiliated cavity position is classified, then the image at each position is divided into clear image and blurred picture, detailed process is as follows:
S ' 2001a, using preset position disaggregated model by the video each frame image and default tag library in not Image with cavity position compares, with the corresponding cavity position of frame image each in the determination video；
Wherein, position disaggregated model can use pattern classification network A lexNet, GoogleNet or other models.Mark The standard picture at each position in cavity is stored in label library.
S ' 2002a, determine that each cavity position is corresponding each in the video using preset two classifier of clarity Frame image is the probability of clear image；If the probability is greater than preset value, it is determined that the frame image is clear image, is otherwise determined The frame image is blurred picture.
Wherein, two classifier of clarity uses convolutional neural networks, the last layer in the structure of the convolutional neural networks For a full articulamentum, after image is inputted the convolutional neural networks, which can export a range in [0,1] Between probability value, when the probability value be greater than preset value, then the image can be classified as clear image, when the probability value is lower than or waits In preset value, then the image can be classified as blurred picture.
It will be appreciated that convolutional neural networks (Convolutional Neural Network, CNN) are a kind of feedforward minds Through network, its artificial neuron can respond the surrounding cells in a part of coverage area, have for large-scale image procossing Outstanding performance, it can specifically include convolutional layer (convolutional layer) and pond layer (pooling layer).
It will be appreciated that so-called clear image and blurred picture are opposite concepts, it is by image as clear figure The probability of picture compared with preset value and determination, wherein the size of preset value and the line of demarcation of clear image and blurred picture have It closes, if preset value is larger, otherwise it can be considered as fuzzy graph that illustrating that only clarity is higher, which just can be considered as clear image, Picture.
For example, first by step S ' 2001a by each frame image in video be divided into m (m be inside cavity position number Amount) a file, the corresponding image in a cavity position is stored in each file, it then will by step S ' 1022a Image in each file is finely divided, and confirms that it belongs to clear image or blurred picture.
It is, of course, also possible to first divide the image into clear image and blurred picture, then cavity position belonging to image is carried out Classification, detailed process is as follows:
S ' 2001b, preset two classifier of clarity is used to determine that each frame image is clear image in the video Probability；If the probability is greater than preset value, it is determined that the frame image is clear image, otherwise determines that the frame image is fuzzy graph Picture；
S ' 2002b, using preset position disaggregated model by each frame clear image and each frame blurred picture respectively with The image of different cavity body region compares in default tag library, with each frame clear image of determination and each frame blurred picture point Not corresponding cavity position.
It will be appreciated that the explanation in relation to content can refer to step S ' 2001a and S ' in S ' 2001b and S ' 2002b 2002a。
For example, each frame image in video two files-are divided by step S ' 2001a first clearly to scheme As file and blurred picture file, then determined corresponding to the image in each file by step S ' 2001b Cavity position.
Both the above mode can determine the corresponding clear image in each position and blurred picture, but latter approach Processing speed it is quicker so that the real-time of the process flow of entire method is much better.
S ' 300, the corresponding blurred picture in each position is carried out at deblurring using generation confrontation network trained in advance Reason obtains corresponding de-blurred image；
That is, fighting network by generation trained in advance carries out deblurring processing to each blurred picture, obtain To the corresponding de-blurred image of this blurred picture.
Wherein, the training process for generating confrontation network includes the following steps:
A1, the first training sample for concentrating preset training sample are separately input into few two different convolutional Neural nets Deblurring processing is carried out in network, obtains at least two frame de-blurred images；Wherein, first training sample is blurred picture；
Wherein, at least two different convolutional neural networks can be selected from numerous convolutional neural networks, for example, super Resolution ratio convolutional neural networks (referred to as SRCNN), efficient sub-pix convolutional neural networks (ESPCN) and be based on Unified Network (unet) at least two in convolutional neural networks.It is, of course, also possible to select other kinds of convolutional neural networks.
For example, as shown in figure 4, the first training sample is inputted SRCNN, ESPCN respectively and based on the convolutional Neural of unet In network, SRCNN can export corresponding de-blurred image, ESPCN is right after carrying out deblurring processing to the first training sample After first training sample carries out deblurring processing, corresponding de-blurred image can be exported, the convolutional neural networks based on unet exist After carrying out deblurring processing to the first training sample, corresponding de-blurred image can be exported.
A2, fusion treatment is carried out at least two frame de-blurred images, obtains that first training sample is corresponding to be gone Blurred picture；
Due to passing through multiple available de-blurred images of step A1, it is therefore desirable to which to this, multiple are gone by the way of certain Blurred picture carries out fusion treatment, obtains a de-blurred image as the final corresponding de-blurred image of the first training sample.
Wherein, there are many modes of fusion treatment, for example, being melted by the way of weighting to multiple de-blurred images It closes.For another example to the mode that multiple de-blurred images are aligned one by one, obtain include multi-layer image stitching image, then Multi-layer image is projected in a certain coordinate system, multiple images are obtained, then this multiple image is carried out by the way of weighting Fusion.
In the case where the quantity of convolutional neural networks is 3, three different convolutional neural networks are to first instruction Practice after sample carries out deblurring processing and obtains the first de-blurred image, the second de-blurred image and third de-blurred image, it is above-mentioned The amalgamation mode of latter citing may include steps of:
A21, the second de-blurred image C2 is directed to the first splicing that the first de-blurred image C1 generates two layers Image M1；
A22, the third de-blurred image C3 is directed at the second spliced map that the first stitching image M1 generates three layers As M2；
A23, described three layers of the second stitching image M2 is projected in global coordinate system, obtains three projected images；
A24, the overlapping region of three images is carried out by pixel fusion using weigthed sums approach, obtains first instruction Practice the corresponding de-blurred image of sample.
It is, of course, also possible to which there are other modes carries out fusion treatment to image, obtains the first training sample and finally correspond to De-blurred image.
A3, the probability that the corresponding de-blurred image of first training sample is clear image is calculated, and according to the probability Judge whether the corresponding de-blurred image of first training sample is clear image:
If so, the training process for generating confrontation network terminates；
Otherwise, the second training sample corresponding de-blurred image of first training sample and the training sample concentrated Originally training is compared, obtains the corresponding de-blurred image of first training sample relative to the total of second training sample Penalty values, according to the total losses value to the deblurring parameter of described at least two different convolutional neural networks and/or described Fusion parameters used by fusion treatment are adjusted, and return step A1.
Wherein, second training sample is clear image and second training sample and first training sample pair Answer the intracorporal same area of the chamber.
Wherein, the calculating process of total losses value may include following process:
A31, the corresponding deblurring figure of first training sample is calculated separately using at least two different loss functions As the penalty values relative to second training sample；
A32, at least two penalty values that described at least two different loss functions are calculated are added up, is obtained Total losses value of the corresponding de-blurred image of first training sample relative to second training sample.
For example, loss function is denoted as d_loss1, d_loss2, d_loss3 respectively, utilizes these three loss functions point It does not calculate penalty values of the corresponding de-blurred image of the first training sample relative to second training sample, then will obtain Three penalty values summations, obtain total losses value.
For example, arbiter to deblurring treated probability that image differentiated between [0,1], if probability Less than determined threshold value, then the clear image gap for representing de-blurred image relative to same position is larger, cannot be by the first training sample This corresponding de-blurred image regards as clear image, thus need in deblurring treatment process deblurring parameter and/or Fusion parameters in fusion process are adjusted, and to reduce total losses value, are then gone using deblurring parameter adjusted Fuzzy Processing and/or use fusion parameters adjusted carry out fusion treatment and obtain the corresponding deblurring of the first training sample again Image, and then judge whether the de-blurred image is clear image again constantly recycles, until obtained de-blurred image can be with Until being considered clear image
Wherein, as shown in figure 5, generate confrontation network may include two sub-networks-deblurring sub-network and differentiate sub-network, Deblurring sub-network therein can execute above-mentioned steps A1 and step A2, differentiate that sub-network can execute above-mentioned steps A3 and step Rapid A4, the process of parameter adjustment are also to adjust the process of the network parameter in deblurring sub-network.Confrontation network is generated as one Kind deep learning model is in recent years for one of the method for unsupervised learning most prospect in complex distributions.Pass through deblurring Sub-network (be referred to as generate model, Generative Model) and differentiation sub-network (it is referred to as discrimination model, Discriminative Model) mutual Game Learning, fairly good output result can be generated.Deblurring sub-network and differentiation Sub-network might not all be neural network, it is only necessary to be that can be fitted the corresponding function generated and differentiate.But practical application In generally use deep neural network as deblurring sub-network and differentiate sub-network, a good generation confrontation network needs Have good training method, otherwise may due to neural network model freedom and cause to export undesirable.
S ' 400, it is synthesized each frame clear image and de-blurred image to obtain the clear of the cavity sequentially in time Clear video, and the clear video is shown.
Since de-blurred image is clear image, sequentially in time by de-blurred image and original clear image After synthesis, the corresponding clear video of original video can be obtained, the clear video be by the fuzzy frame in original video into Become articulating frame after the certain processing of row, so that entire video is all than more visible.
Here, processing unit removes the blurred picture in the collected video of cavity mirror system by generating confrontation network Then the clear image obtained after original clear image and deblurring processing is carried out synthesizing sequentially in time by Fuzzy Processing To clear video, the problem of image blurring as caused by equipment shake under adverse circumstances can be solved, effectively to perform the operation or checking Hysteroscope video, the accuracy of raising target positioning mitigate the diagnosis of doctor or burden of performing the operation in real time and clearly for process offer.
(3.2) processing unit is also used to execute following methods, to realize that dynamic object positions:
S100, the time according to frame image each in the video and color, select key frame images from the video；
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, key frame images can be extracted 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.
S200-, the edge black surround of the key frame images is smoothly located according to the visual field parameter of the hysteroscope camera lens After reason is filtered denoising to the image after smoothing processing using high-pass filter, and use median filter is to filtering and noise reduction Image be 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.
Here it is provided with step S200- in step S200, which mainly realizes the optimization processing to key frame images, this For realizing general object of the present invention it is not necessary to, therefore in certain embodiments can not include step S200-。
S200, the YOLO target detection model that each key frame images input is preset to training, obtain multiple with target The image of posting and target category mark；
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 according to It can so learn to correspond to { w1, w2, w3...wn}={ 0,0,0 ... 1 } good, old maximum pond, and subsequent layer can To obtain more information, gradient flows through all values in one layer when backpropagation.Sort-pooling can be realized faster more It restrains well, the Optimized Iterative time retains more image informations, while also protruding important image information, so that mesh It demarcates position and identifies that more accurate and treatment effeciency is 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.
In this step, the image identified with target posting and target category is determined using YOLO target detection model Process can specifically include:
S201, each key frame images are divided into S*S grid, s is the integer greater than 1；
S202, 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；
S203, 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；
S204, 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.
S300, multiple described images identified with target posting and target category are synthesized, it is fixed obtains target Digital video, and the target positioning video is shown.
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.
Here, processing unit obtains hysteroscope video first, then extracts key frame images therein, then using preparatory Trained YOLO target detection model position to the target in key frame images and determines target type, then will have target Posting and the image of target category mark are synthesized, and dynamic target positioning video is obtained.Due to YOLO trained in advance Including being clustered using K-centers clustering method to training sample data collection in the training process of target detection model, and It can be effectively improved " noise " tender subject by the way of K-centers cluster, so as to improve target positioning video Image quality.Identification of the present invention due to carrying out target positioning and target type using target detection model simultaneously, processing effect Rate is high, processing speed is fast, can accomplish real-time target positioning and target type discrimination.
, can also it will be appreciated that although above-mentioned target position fixing process is that the video being converted to for video camera carries out To be carried out for clear video obtained in (3.1), that is to say, that processor first carries out the video that video camera is converted to Processing obtains clear video, and target positioning, the target positioning video obtained in this way are then carried out on the basis of clear video Be clearly, and in medical staff refer to.
It will be appreciated that the above method performed by processing unit is realized based on image processing techniques, mesh therein Mark can according to need setting, for example, the intracorporal some abnormal conditions of chamber, can be identified different by above-mentioned object localization method The position of reason condition and the type of abnormal conditions.
It will be appreciated that above-mentioned processing unit is the device independently of incorporating chamber lens device, the particular hardware of the device It may include processor, memory and display, be stored with computer program in memory, when these computer programs are performed When may be implemented the above method, display executes clear video and/or target positioning video obtained by the above method for showing. The device is connect with the video camera in incorporating chamber lens device, obtains the video that video camera is converted to, and carry out based on video Then image procossing is shown with obtaining clear video and/or target positioning video, for medical staff's reference.
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 intelligent integral cavity mirror system, which is characterized in that the system comprises incorporating chamber lens device and processing unit, Wherein:
The incorporating chamber lens device includes the central control unit of multiple functional modules, the multiple functional module work of control The power supply device of power supply is provided with for the multiple functional module；The multiple functional module includes the gas for inflating to cavity Abdomen instrument, human-computer interaction screen, cold light source and video camera, the pneumoperitoneum instrument, the human-computer interaction screen, the cold light source and the camera shooting Machine is connected to the central control unit；The cold light source and the video camera are all connected with endoscope optical；The cold light source Light source is provided for the endoscope optical；The collected optical signal of the endoscope optical is converted to video by the video camera, And the video is sent to the human-computer interaction screen and carries out the displaying of hysteroscope video；Wherein:
The pneumoperitoneum instrument includes proportional control valve, switch electromagnetic valve, gas bleeder valve, mass-air-flow sensor and baroceptor, in which: The proportional control valve, the switch electromagnetic valve, the gas bleeder valve, the mass-air-flow sensor and the baroceptor are all provided with It sets on supply air line, and is connect with the central control unit；The mass-air-flow sensor is for detecting the air supply pipe The throughput parameter on road；The baroceptor is used to detect the pneumatic parameter of the supply air line；Wherein, the supply air line The pipeline of gas is inputted to patient's body lumen for the pneumoperitoneum instrument；The central control unit for obtain the throughput parameter and The pneumatic parameter according to the throughput parameter and is preset when the pneumatic parameter is less than preset first pneumatic parameter Standard Gases flow parameter output pwm signal, the pwm signal is for adjusting the switch time ratio of the control valve Section is adjusted the throughput parameter in the supply air line with realizing；It is also used to be greater than or equal in the pneumatic parameter When first pneumatic parameter, the proportional control valve and the switch electromagnetic valve are closed；It is also used to big in the pneumatic parameter When the second pneumatic parameter, gas bleeder valve is opened；Wherein, second pneumatic parameter is greater than first pneumatic parameter；
The power supply device includes Switching Power Supply and multiple electromagnetic interference suppression circuits；The Switching Power Supply includes sequentially connected First rectification circuit, transformer and multiple second rectification circuits；The alternating current that first rectification circuit is used to be inputted turns Unidirectional voltage needed for becoming the transformer；The transformer includes an armature winding and multiple secondary windings, described first Grade winding is connect with the output end of first rectification circuit, the multiple secondary windings and the multiple second rectification circuit Input terminal connects one to one；Each second rectification circuit is used to the output voltage for corresponding to secondary windings being changed into corresponding function DC voltage needed for energy module；The multiple electromagnetic interference suppression circuit, input terminal and the multiple second rectification circuit Output end connect one to one, output end is used to connect one to one with the input terminal of the multiple functional module, is used for The bi-directional electromagnetic between the multiple functional module and the Switching Power Supply is inhibited to interfere；Wherein, the secondary windings, described The quantity of two rectification circuits, the electromagnetic interference suppression circuit and the functional module is all the same；
The processing unit is used to execute: S ' 100, obtaining the video that the video camera is converted to；S ' 200, the view is determined The corresponding cavity position of each frame image and classify each frame image to obtain the cavity according to readability in frequency Clear image corresponding to interior different parts and blurred picture；S ' 300, network is fought to each using generation trained in advance The corresponding blurred picture in position carries out deblurring and handles to obtain corresponding de-blurred image；S ' 400, it by each frame clear image and goes Blurred picture is synthesized to obtain the clear video of the cavity sequentially in time, and the clear video is shown；
The processing unit is also used to execute: S100, according to the time of each frame image in the video or the clear video and Color selects key frame images from the video or the clear video；S200, each key frame images input is preset Trained YOLO target detection model obtains multiple images identified with target posting and target category；S300, will be described Multiple are synthesized with images that target posting and target category identify, and obtain target positioning video, and by the target Positioning video is shown；Wherein, the training process of the YOLO target detection model includes at least: poly- using K-centers Class method clusters training sample data collection.
2. system according to claim 1, which is characterized in that
The 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 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；And/or
The 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, the camera shooting master Plate is used to the electric signal being converted to video, and the video is sent to the human-computer interaction screen and the processing unit.
3. system according to claim 1, which is characterized in that
The incorporating chamber lens device further include: storage unit connects the central control unit, for storing multiple age brackets Corresponding data set；
It is corresponding, the multiple age bracket options of human-computer interaction screen display, and the human-computer interaction screen is also used to detecting When any option in the multiple age bracket option is selected, trigger signal is sent to the central control unit；
The central control unit is also used to when receiving the trigger signal, according to the corresponding age bracket of the option selected Corresponding data set determines the air pressure recommendation and throughput recommendation for being directed to the age bracket, and by the air pressure recommendation It is shown on the human-computer interaction screen with the throughput recommendation.
4. system according to claim 3, which is characterized in that
The pneumoperitoneum instrument further include: temperature sensor is arranged on the supply air line, connect with the central control unit, For detecting the gas temperature in the supply air line；
The corresponding data set of all age group stored in the storage unit includes a plurality of record content, in every record Hold includes air pressure, throughput, the access times under the gentle flow set of the air pressure, abdominal cavity Pressure Inflation to specified air pressure threshold value Average time used and this note within the scope of gas mean temperature, abdominal cavity Pressure Inflation to specified air pressure threshold value when in range Record the timestamp of content；
Corresponding, the central control unit is used to determine that the air pressure recommendation and the throughput are recommended using following steps Value:
The suitable score value of every record content is calculated using the first formula, first formula includes:
In formula, g is to be suitble to score value, a1、a2And a3For weighted value, tsmaxIt is maximum for access times in a plurality of record content Value, tmaxFor time used average in a plurality of record content maximum value, c is to record abdominal cavity air pressure in content in this Gas mean temperature when being inflated within the scope of specified air pressure threshold value, t are to record abdominal cavity Pressure Inflation in content in this extremely to refer to Determine the average time used within the scope of air pressure threshold value, ts is to record the access times in content in this；
Air pressure in record content corresponding to maximum value in obtained multiple suitable score values is recommended as the air pressure It is worth and using the throughput in the corresponding record content of the maximum value as throughput recommendation.
5. system according to claim 4, which is characterized in that the central control unit is also used to the storage unit In data set be updated, specifically include:
The pneumoperitoneum instrument in use, maintained in specified threshold in first time abdominal cavity air pressure and the difference of default pressure When preset duration, to current gas pressure, current throughput, current gas temperatures, this abdominal cavity Pressure Inflation to specified air pressure threshold value Time used and timestamp in range are recorded；
If in current record content including the record of the current gas pressure and the current throughput, to corresponding using secondary Number plus 1, and timestamp is updated, updated average time used and gas mean temperature are calculated using the second formula, Second formula includes:
In formula, n is updated average time used or updated gas mean temperature, and o is average used before updating Gas mean temperature before time or update, ts are access times before this update, and r is that this abdominal cavity Pressure Inflation extremely refers to Determine the time used or the current gas temperatures within the scope of air pressure threshold value.
6. system according to claim 1, which is characterized in that each electromagnetic interference suppression circuit includes sequentially connected Schottky diode, differential mode LC filter unit and common mode LC filter unit；Wherein:
The anode of the Schottky diode connects the positive output end of corresponding second rectification circuit；
The differential mode LC filter unit includes differential mode inductance and at least two first capacitors in parallel；One end of the differential mode inductance The cathode of the Schottky diode is connected, the another of the differential mode inductance is arranged in described at least two first capacitors in parallel Between end and the negative output terminal of the second rectification circuit of the correspondence, the negative output terminal of second rectification circuit of correspondence is grounded；
The common mode LC filter unit includes common mode inductance and at least two the second capacitors in parallel, one in the common mode inductance One end of winding connects the other end of the differential mode inductance, and one end of another winding connects the correspondence in the common mode inductance A winding of the common mode inductance is arranged in the negative output terminal of second rectification circuit, described at least two second capacitors in parallel The other end and the common mode inductance another winding the other end between；The other end of one winding of the common mode inductance As the positive output end of the electromagnetic interference suppression circuit, which is used to connect the positive input terminal of corresponding function module；Institute Negative output terminal of the other end as the electromagnetic interference suppression circuit of another winding of common mode inductance is stated, which is used for Connect the negative input end of corresponding function module.
7. system according to claim 1, which is characterized in that the processing unit is according to the video or the clear view The time of each frame image and color in frequency select the process packet of key frame images from the video or the clear video It includes:
S201, using the first frame image in the video or the clear 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 or the clear 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 or the clear 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 or the clear video, si=s (ti, ci), tiIt is the i-th frame image in institute State time point locating in video or the clear video, ciFor the color matrix of the i-th frame image.
8. system according to claim 1, which is characterized in that include in the network structure of the YOLO target detection model Pond layer, the pond layer can successively sort n activation primitive value from small to large, from small to large successively by n weighted value N weighted value is multiplied with corresponding activation primitive value by sequence respectively, calculates the average value of n multiplied result, and will be described flat Mean value is as final activation primitive value.
9. system according to claim 1, which is characterized in that the processing unit is also used to training in advance and obtains the life At confrontation network, detailed process includes:
A1, the first training sample for concentrating preset training sample are separately input into few two different convolutional neural networks Deblurring processing is carried out, at least two frame de-blurred images are obtained；Wherein, first training sample is blurred picture；
A2, fusion treatment is carried out at least two frame de-blurred images, obtains the corresponding deblurring of first training sample Image；
A3, the probability that the corresponding de-blurred image of first training sample is clear image is calculated, and according to the probabilistic determination Whether the corresponding de-blurred image of first training sample is clear image:
If so, the training process for generating confrontation network terminates；
Otherwise, the second training sample corresponding de-blurred image of first training sample and the training sample concentrated into Row comparative training obtains total losses of the corresponding de-blurred image of first training sample relative to second training sample Value, according to the total losses value to the deblurring parameter and/or the fusion of described at least two different convolutional neural networks Fusion parameters used by handling are adjusted, and return step A1；Wherein, second training sample is clear image and institute It states the second training sample and first training sample corresponds to the intracorporal same area of the chamber.
10. system according to claim 9, which is characterized in that the quantity of the convolutional neural networks is 3, and three are not With convolutional neural networks first training sample carried out obtaining the first de-blurred image after deblurring processing, second remove mould Paste image and third de-blurred image；
Corresponding, the processing unit carries out fusion treatment at least two frame de-blurred images, comprising: goes described second Blurred picture is directed at the first stitching image that first de-blurred image generates two layers；The third de-blurred image is aligned First stitching image generates three layers of the second stitching image；Described three layers of the second stitching image is projected into world coordinates In system, three projected images are obtained；The overlapping region of three images is carried out by pixel fusion using weigthed sums approach, is obtained The corresponding de-blurred image of first training sample.
CN201810997377.7A 2018-08-29 2018-08-29 Intelligent integrated endoscope system Active CN109124782B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810997377.7A CN109124782B (en) 2018-08-29 2018-08-29 Intelligent integrated endoscope system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810997377.7A CN109124782B (en) 2018-08-29 2018-08-29 Intelligent integrated endoscope system

Publications (2)

Publication Number Publication Date
CN109124782A true CN109124782A (en) 2019-01-04
CN109124782B CN109124782B (en) 2020-09-22

Family

ID=64829202

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810997377.7A Active CN109124782B (en) 2018-08-29 2018-08-29 Intelligent integrated endoscope system

Country Status (1)

Country Link
CN (1) CN109124782B (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109243586A (en) * 2018-08-31 2019-01-18 合肥德铭电子有限公司 Configure the hysteroscope intelligence edge calculations system of central control module
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
CN111759455A (en) * 2020-06-29 2020-10-13 上海交通大学 Bubble suction instrument for electric excision operation
CN114972349A (en) * 2022-08-01 2022-08-30 山东西曼克技术有限公司 Carrier roller running state detection method and system based on image processing

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN203352463U (en) * 2013-07-17 2013-12-18 浙江沃尔德电力电子有限公司 multi-channel isolated output and anti-interference intelligent power capacitor
CN204012697U (en) * 2014-09-01 2014-12-10 成都乐创自动化技术股份有限公司 A kind of safeguard structure that is applied to laser machine internal control part power supply
CN105431093A (en) * 2013-08-06 2016-03-23 奥林巴斯株式会社 Pneumoperitoneum apparatus
CN107730458A (en) * 2017-09-05 2018-02-23 北京飞搜科技有限公司 A kind of fuzzy facial reconstruction method and system based on production confrontation network
CN108009473A (en) * 2017-10-31 2018-05-08 深圳大学 Based on goal behavior attribute video structural processing method, system and storage device
US20180225588A1 (en) * 2017-02-03 2018-08-09 Adxcel Inc. Accelerated machine learning optimization strategy to determine high performance cluster with minimum resources
CN108416752A (en) * 2018-03-12 2018-08-17 中山大学 A method of image is carried out based on production confrontation network and removes motion blur

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN203352463U (en) * 2013-07-17 2013-12-18 浙江沃尔德电力电子有限公司 multi-channel isolated output and anti-interference intelligent power capacitor
CN105431093A (en) * 2013-08-06 2016-03-23 奥林巴斯株式会社 Pneumoperitoneum apparatus
CN204012697U (en) * 2014-09-01 2014-12-10 成都乐创自动化技术股份有限公司 A kind of safeguard structure that is applied to laser machine internal control part power supply
US20180225588A1 (en) * 2017-02-03 2018-08-09 Adxcel Inc. Accelerated machine learning optimization strategy to determine high performance cluster with minimum resources
CN107730458A (en) * 2017-09-05 2018-02-23 北京飞搜科技有限公司 A kind of fuzzy facial reconstruction method and system based on production confrontation network
CN108009473A (en) * 2017-10-31 2018-05-08 深圳大学 Based on goal behavior attribute video structural processing method, system and storage device
CN108416752A (en) * 2018-03-12 2018-08-17 中山大学 A method of image is carried out based on production confrontation network and removes motion blur

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
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》 *
黄小天 等: "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》 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109243586A (en) * 2018-08-31 2019-01-18 合肥德铭电子有限公司 Configure the hysteroscope intelligence edge calculations system of central control module
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
CN111759455A (en) * 2020-06-29 2020-10-13 上海交通大学 Bubble suction instrument for electric excision operation
CN114972349A (en) * 2022-08-01 2022-08-30 山东西曼克技术有限公司 Carrier roller running state detection method and system based on image processing
CN114972349B (en) * 2022-08-01 2022-10-25 山东西曼克技术有限公司 Carrier roller running state detection method and system based on image processing

Also Published As

Publication number Publication date
CN109124782B (en) 2020-09-22

Similar Documents

Publication Publication Date Title
CN109124782A (en) Intelligent integral cavity mirror system
CN109171605A (en) Intelligent edge calculations system with target positioning and hysteroscope video enhancing processing function
CN109124702A (en) Configure the intelligent cavity mirror system of pneumoperitoneum control and central control module
CN105050247B (en) Light intelligent regulating system and its method based on expression Model Identification
CN109243586A (en) Configure the hysteroscope intelligence edge calculations system of central control module
CN111079833B (en) Image recognition method, image recognition device and computer-readable storage medium
CN109223178A (en) Hysteroscope intelligence edge calculations system with target positioning function
Mao et al. A brain–robot interaction system by fusing human and machine intelligence
CN108765611A (en) A kind of dynamic human face identification Work attendance management system and its management method
CN109350239A (en) Intelligent integral robot cavity mirror system with target positioning function
CN109241898A (en) Object localization method and system, the storage medium of hysteroscope video
CN109044272A (en) Sleep state monitoring and sleep environment parameter regulation method, system and terminal device
CN104994347A (en) Intelligent security video monitoring system and detection processing method thereof
CN108926319A (en) Configure the intelligent movable integration cavity mirror system of high definition video processing module
CN106851229A (en) A kind of method and system of the security protection intelligent decision based on image recognition
CN105934028A (en) Adaptive illumination adjustment system and medical helmet
CN107527025A (en) A kind of recognition of face intelligent monitor system
CN109363613A (en) Inhibit the intelligent movable integration cavity mirror system of function with electromagnetism
CN109363614A (en) Intelligent integral robot cavity mirror system with high definition video enhancing processing function
CN109171606A (en) Intelligent integral robot cavity mirror system
CN107949121A (en) One kind is used to build high-efficiency and energy-saving type intelligent lighting system and its method of work
CN110378426A (en) A kind of Breast Nodules aided diagnosis method and system based on YOLOV3
CN109151275A (en) Intelligent edge calculations system with real-time hysteroscope video enhancing processing function
CN114663293A (en) Image enhancement method and device, electronic equipment and endoscope system
CN109288487A (en) Inhibit the intelligent cavity mirror system of function with the processing of high definition video and electromagnetism

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