CN105979201A - Intelligent wearable device based on parallel processor - Google Patents
Intelligent wearable device based on parallel processor Download PDFInfo
- Publication number
- CN105979201A CN105979201A CN201610216172.1A CN201610216172A CN105979201A CN 105979201 A CN105979201 A CN 105979201A CN 201610216172 A CN201610216172 A CN 201610216172A CN 105979201 A CN105979201 A CN 105979201A
- Authority
- CN
- China
- Prior art keywords
- intelligent wearable
- parallel processor
- input
- microprocessor
- display
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/06—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
- G06N3/061—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons using biological neurons, e.g. biological neurons connected to an integrated circuit
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Theoretical Computer Science (AREA)
- Molecular Biology (AREA)
- Neurology (AREA)
- Computational Linguistics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Microelectronics & Electronic Packaging (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to an intelligent wearable device based on a parallel processor. The intelligent wearable device comprises an image receiving module, a control module, a parallel computation coprocessor, an external input and output device and an external nonvolatile storage device, and is characterized in that the image receiving module is used for receiving image data through a wireless mode; the control module is composed of a microprocessor or a microcontroller, a frame buffer, a memory unit and various types of interfaces, the microprocessor or the microcontroller is used for generating control information required by the image receiving module and a micro-display, and is also used for managing operations of the parallel computation processor and processing data transmitted by the parallel computation processor at the same time; the parallel computation processor is used for analyzing and processing data transmitted by the external input and output device; and the external input and output device comprises the micro-display, a camera and other input and output devices. The system uses an artificial neural network structure to act as a data analysis and recognition module, thereby enhancing the data processing capacity, reducing the load of the microprocessor, enabling the device to have high intelligence.
Description
Technical field
The present invention designs a kind of intelligent wearable device based on parallel processor, including image receiver module, control module, concurrent operation coprocessor, outside input-output equipment.
Background technology
Micro-display (is the display of a kind of more specific form, generally Diagonal Dimension is called miniscope less than the display of 3.3cm.Its own physical size is the least, and pixel is away from general only ten microns, but can carry out large screen display.Micro-display is widely used in various consumer electronics, such as minitype projection machine, wear-type 3D high definition micro-display, wear-type medical system, and military affairs can be applicable to air force pilot's headgear system, ground force's individual soldier's Integrated Helmet System.
Artificial neural network is to use for reference biological neural network structure and biological neuron working mechanism, simulate the information processing system of some function of human brain to a certain extent, i.e. simulate some basic function of biological neuron, build the artificial neuron with independent processing ability, by artificial neural network is trained, adjust interconnected relationship between a large amount of neuron of network internal, so that it may so that Network Recognition input and the mapping relations of output, reach the purpose of information processing.Artificial neural network now has been applied to many fields, such as fields such as system identification (Aero-Space, power system etc.), pattern recognition (speech recognition, character recognition etc.), Based Intelligent Control.Artificial neural network mainly has software and hardware two kinds to realize type, software realizes depending on computer, owing to the operand of neutral net is the biggest, computing power is required the highest, and software approach can not be real realize parallel processing, so for the high application of some requirement of real-times, it is necessary to use hardware circuit to realize, to complete processing the most in real time of data.
At present, Chinese patent, its Patent No. 201210196780.2 " wireless micro-display (41) system ", this system achieves wireless receiving and dispatching and the display of image.The deficiency that this system exists is: transmitting-receiving view data that can only be simple and display image, image receiver module can not the information such as active obtaining external image and sound, can not be analyzed view data identifying.
Summary of the invention
Present invention aims to the defect that prior art exists, it is provided that a kind of intelligent wearable device based on parallel processor.This system utilizes the external equipment such as photographic head (43), sound receiver (44) actively to accept external information, and utilizing artificial neural network that the information received is analyzed, processes, is identified, output result is in the upper display of micro-display (41).
According to above-mentioned inventive concept, the present invention uses following technical scheme:
A kind of intelligent wearable device based on parallel processor, including image receiver module, control module, concurrent operation coprocessor, outside input-output equipment, external non-volatile storage device, it is characterized in that: described image receiver module connects external non-volatile storage device and control module, control module connects concurrent operation coprocessor and outside input-output equipment；Described image receiver module is for wirelessly receiving view data from image transmitting terminal；Described control module is made up of microprocessor or microcontroller, frame buffer, internal storage location and each class interface, microprocessor or microcontroller, for producing image receiver module and the control information of micro-display needs, are the most also used for managing the work of concurrent operation coprocessor and processing its data transmitted；Concurrent operation coprocessor is for the Data Analysis Services transmitting external input device;Outside input-output equipment includes micro-display and photographic head, is used for showing image, obtaining external information；External non-volatile storage device is used for the program that stores, data and parameter.
In above-mentioned intelligent wearable device based on parallel processor based on stochastic neural net, described concurrent operation coprocessor (3) can use neural network structure, connects outside input-output equipment (4), microprocessor or microcontroller (21) and external non-volatile storage device (5).It is used for analyzing and identifying the information that external input device (such as photographic head (43)) inputs, and in micro-display (41) these information of upper display.Described neutral net includes input layer (A), intermediate layer (B) and output layer (C), and wherein intermediate layer (B) is if can be made up of dried layer.Described input layer (A) is made up of I input neuron, and described input neuron (A) accepts the first data (AI), exports the first data sequence (AO)；(B) can have some sublayers in described intermediate layer, each sublayer is made up of Jn relay cell, described intermediate layer neuron (B) accepts the first data sequence (AO), argument sequence (R), complete RBF computing, export the second data sequence (BO)；Described output module is made up of K output neuron, and described output neuron accepts the second data sequence (BO), argument sequence (R), exports the 3rd data sequence (CO), wherein I, and Jn, K are the integer more than or equal to 1；The outfan of input neuron is connected to the input of relay cell, and the outfan of relay cell is connected to the input of output neuron, and it connects can use complete being connected or the most connected.Additionally, parameter required for each layer of neutral net is stored in external non-volatile storage device (5), external non-volatile storage device (5) can be nonvolatile memory such as FLASH, and the interface of this storage device can be IDE or SATA.
Described neural network structure can be made up of stochastic neural net based on probabilistic operation.This neutral net by binary number by being converted to random data sequence, so that sequence occurring, the probability of 0 or 1 represents the numerical value used in deterministic parameters calculation, multiplying is completed with logical AND gate or logic XOR gate, exponent arithmetic is completed with one-dimensional linear state machine or two-dimensional matrix state machine, after network completes to calculate, then probit is converted to straight binary number.Described each data sequence is to occur in data sequence in a period of time that the probit of 0 or 1 represents.Each argument sequence is converted to random parameter sequence by random transition device (31).Described relay cell is all with the data sequence of above one layer of output for inputting independent variable, with relevant parameter sequence as function parameter, complete RBF computing, calculating process all uses probability number (i.e. occurring in data sequence in a period of time that the probability of 0 or 1 represents a numerical value), exports random data sequence.This stochastic neural net has the ability of process in real time.
Described RBF kind includes but not limited to Gaussian function, many quadratic functions, inverse many quadratic functions, thin plate spline function, cubic function, linear function.Neutral net includes but not limited to stochastic neural net, semi-random neutral net, BP neutral net, PNN neutral net, Hopfield network, SOM neutral net, LVQ neutral net.
In above-mentioned intelligent wearable device based on parallel processor based on stochastic neural net, described microprocessor or microcontroller (21) can use embedded microprocessor, it is connected with image receiver module (1), concurrent operation coprocessor (3), frame buffer (22), internal storage location (23), external non-volatile storage device (5), micro-display (41) and outside input-output equipment (4), is used for producing the work of a series of control information management each equipment.This microprocessor connects the outside input of button, it is simple to extraneous input control information；Whether control image receiver module (1) accepts to open wireless receiving end (111) accepts view data；Control stochastic neural net whether to work, how to process the output result of stochastic neural net；Control frame buffer (22) carries out writing view data respectively, reads manipulation of image data；This microprocessor is connected with micro-display (41), is used for controlling the resolution of micro-display (41), gray scale, brightness, contrast, color saturation, Gamma correction, picture format, video standard, scan mode, input mode；Control the work of other external input device.External non-volatile storage device (5) can store corresponding control program, driver and data.
In above-mentioned intelligent wearable device based on parallel processor based on stochastic neural net, described picture receiver (11) is connected image decoder module (112) by wireless receiving end (111) and constitutes, image decoder module (112) is used for decompressed image data, and view data is delivered to frame buffer (22).Described wireless receiving end is linked in sequence is constituted by tuner (1111), low-noise amplifier (1112), frequency mixer (1113), radio-frequency tuner (1114), digital demodulation channel-decoding (1115) and demultiplexer (1116).Described picture decoding is linked in sequence by entropy code (1121), inverse quantizer (1122) and IDCT (1123), and has table specification (1124) to connect entropy code (1121) and table specification (1125) connection inverse quantizer (1122) composition.
View data is sent to image receiver module (1) by image sending module (6) by wireless communication transmissions agreement；Described wireless communication protocol is UWB, WHDI, WirelessHD, WIFI, WiMAX, WiDi or LTE；
In above-mentioned intelligent wearable device based on parallel processor based on stochastic neural net, described frame buffer (22) is random access memory, and described frame buffer (22) can be SRAM SRAM or dynamic RAM DRAM.The view data that it decodes through image receiver module (1) for storage, and the view data transmission that can it be stored shows to micro-display (41).After the view data that image transmitting terminal sends over is stored by described frame buffer (22), when needing repeatedly to show this view data, image transmitting terminal need not send again, as long as directly reading from frame buffer (22).
In above-mentioned intelligent wearable device based on parallel processor based on stochastic neural net, described micro-display (41) is a kind of miniature display device: silicon based LCD micro-display, silicon-based organic light-emitting micro-display, silicon-based inorganic luminescence micro-display, non-silicon-based luminescence micro-display or miniature CRT monitor, Diagonal Dimension is less than 3.3cm, pixel resolution is not less than 640*480, the gray scale of single color component is not less than 32 grades, support binocular 3D application, support brightness, contrast and color saturation adjust, support Gamma correction, support monochrome/YCbCr/RGB picture format, support PAL/NTSC/SMPTE video standard, support line by line or interlaced scan mode, support the input modes such as work VGA/DVI/RGB.It is received in control module (2) such as VGA, DVI, RGB, HDMI, MIPI, MDDI, other kinds of LVDS, SPI, IIC, USB, other kinds of serial ports, parallel port etc. by corresponding video coffret.
In above-mentioned intelligent wearable device based on parallel processor based on stochastic neural net, if described micro-display (41) supports fractal scanning function, the most described micro-display (41) interface answers integrated fractal scanning control module, in order to produce fractal scanning control signal, form more than 256 grades gray-scale Control signals and drive micro-display (41).
In above-mentioned intelligent wearable device based on parallel processor based on stochastic neural net, described output peripheral equipment can be photographic head (43), sound receiver (44), inertial sensor (45) etc., they access in corresponding interface (M), it is eventually connected to artificial neural network and microprocessor or microcontroller (21), for artificial neural network analysis identification, their work is by microprocessor or microprocessor controls.
The present invention compared with prior art, has and the most obviously highlights substantive distinguishing features and marked improvement:
Each equipment work in the system uses embedded microprocessor control system, it is provided that multiple external apparatus interface (M) (such as photographic head (43) etc.), makes the information such as reception equipment energy active obtaining external image；The system uses stochastic neural net structure as data analysis identification module, strengthen data-handling capacity, alleviate microprocessor burden, make equipment be provided with higher intelligent.
Accompanying drawing explanation
Fig. 1 is based on parallel processor the intelligent wearable device structural representation of the present invention.
Fig. 2 is based on parallel processor the intelligent wearable device artificial neural network configuration diagram of a preferred embodiment of the present invention two.
Fig. 3 is based on parallel processor the intelligent wearable device stochastic neural net configuration diagram of a preferred embodiment of the present invention three.
Fig. 4 is based on parallel processor the intelligent wearable device control module configuration diagram of a preferred embodiment of the present invention four.
Fig. 5 is the structural representation of a kind of image sending module of a preferred embodiment of the present invention five.
Fig. 6 is based on parallel processor the intelligent wearable device image receiver module structural representation of a preferred embodiment of the present invention six.
Fig. 7 is based on parallel processor the intelligent wearable device frame buffer SRAM structural representation of a preferred embodiment of the present invention eight.
Fig. 8 is based on parallel processor the intelligent wearable device frame buffer DRAM structural representation of a preferred embodiment of the present invention nine.
Specific embodiment
Below in conjunction with accompanying drawing, the technical scheme of the preferred embodiments of the present invention is further described:
Embodiment one:
With reference to Fig. 1.This intelligent wearable device based on parallel processor, including image receiver module (1), control module (2), concurrent operation coprocessor (3), outside input-output equipment (4).It is characterized in that: described image receiver module (1) connects external non-volatile storage device (5) and control module (2), control module (2) connects concurrent operation coprocessor (3) and outside input-output equipment (4)；Described image receiver module (1) is for wirelessly receiving view data from image transmitting terminal (6)；Described control module (2) is made up of microprocessor, frame buffer (22), internal storage location (23) and external non-volatile storage device (5), the control information that microprocessor or microcontroller (21) are used for producing image receiver module (1) and micro-display (41) needs, is the most also used for managing the work of concurrent operation coprocessor (3) and processing its data transmitted；Concurrent operation coprocessor (3) is for the data transmitting external input device;Outside input-output equipment (4) includes micro-display (41) and photographic head (43) and other input-output equipment, is used for showing image, obtaining external information etc..
Embodiment two:
The present embodiment is essentially identical with embodiment one, is particular in that:
With reference to Fig. 2.In above-mentioned intelligent wearable device based on parallel processor based on stochastic neural net, described concurrent operation coprocessor (3) can use neural network structure, connects outside input-output equipment (4), microprocessor or microcontroller (21) and external non-volatile storage device (5).It is used for analyzing and identifying the information that external input device (such as photographic head (43)) inputs, and in micro-display (41) these information of upper display.Described neutral net includes input layer (A), intermediate layer (B) and output layer (C), and wherein intermediate layer (B) is if can be made up of dried layer.Described input layer (A) is made up of I input neuron, and described input neuron (A) accepts the first data (AI), exports the first data sequence (AO)；(B) can have some sublayers in described intermediate layer, each sublayer is made up of Jn relay cell, described intermediate layer neuron (B) accepts the first data sequence (AO), argument sequence (R), complete RBF computing, export the second data sequence (BO)；Described output module is made up of K output neuron, and described output neuron accepts the second data sequence (BO), argument sequence (R), exports the 3rd data sequence (CO), wherein I, and Jn, K are the integer more than or equal to 1；The outfan of input neuron is connected to the input of relay cell, and the outfan of relay cell is connected to the input of output neuron, and it connects can use complete being connected or the most connected.Additionally, parameter required for each layer of neutral net is stored in external non-volatile storage device (5), external non-volatile storage device (5) can be nonvolatile memory such as FLASH, and the interface of this storage device can be IDE or SATA.
Embodiment three:
The present embodiment is essentially identical with embodiment one, is particular in that:
With reference to Fig. 3.Described neural network structure can be made up of stochastic neural net based on probabilistic operation.This neutral net by binary number by being converted to random data sequence, so that sequence occurring, the probability of 0 or 1 represents the numerical value used in deterministic parameters calculation, multiplying is completed with logical AND gate or logic XOR gate, exponent arithmetic is completed with one-dimensional linear state machine or two-dimensional matrix state machine, after network completes to calculate, then probit is converted to straight binary number.Described each data sequence is to occur in data sequence in a period of time that the probit of 0 or 1 represents.Each argument sequence is converted to random parameter sequence by random transition device (31).Described relay cell is all with the data sequence of above one layer of output for inputting independent variable, with relevant parameter sequence as function parameter, complete RBF computing, calculating process all uses probability number (i.e. occurring in data sequence in a period of time that the probability of 0 or 1 represents a numerical value), exports random data sequence.Additionally, parameter required for each layer of neutral net is stored in external non-volatile storage device (5), external non-volatile storage device (5) can be nonvolatile memory such as FLASH, and the interface of this storage device can be IDE or SATA.
Embodiment four:
The present embodiment is essentially identical with embodiment one, is particular in that:
With reference to Fig. 4.In above-mentioned intelligent wearable device based on parallel processor based on stochastic neural net, described microprocessor can use embedded microprocessor, it is connected with image receiver module (1), concurrent operation coprocessor (3), frame buffer (22), internal storage location (23), external non-volatile storage device (5), micro-display (41) and outside input-output equipment (5), is used for producing the work of a series of control information management each equipment.This microprocessor connects the outside input of button, it is simple to extraneous input control information；Whether control image receiver module (1) accepts to open wireless receiving end (111) accepts view data；Control stochastic neural net whether to work, how to process the output result of stochastic neural net；Control frame buffer (22) carries out writing view data respectively, reads manipulation of image data；This microprocessor is connected with micro-display (41), is used for controlling the resolution of micro-display (41), gray scale, brightness, contrast, color saturation, Gamma correction, picture format, video standard, scan mode, input mode；Control the work of other external input device.External non-volatile storage device (5) can store corresponding control program, driver and data.
Embodiment five:
The present embodiment is essentially identical with embodiment one, is particular in that:
With reference to Fig. 5.In above-mentioned base intelligent wearable device based on parallel processor based on stochastic neural net, a kind of image sending module (6) is as shown in the figure, connected wireless transmission end (62) by picture coding (61) to constitute, picture coding (61) is used for compressing image data, and view data is sent to image receiver module (1) by wireless transmission end (62) by wireless communication transmissions agreement.Described picture coding (61) is linked in sequence by FDCT (611), quantizer (612) and entropy code (613), and has table specification (614) and table specification (615) to connect quantizer (61) and entropy code (613) composition respectively.Described wireless transmission end is linked in sequence by digital frequency synthesizer (621), multiplexer (622), chnnel coding constitutes with manipulator (623), phaselocked loop (624) and radio-frequency channel (625), and described wireless communication protocol is UWB, WHDI, WirelessHD, WIFI, WiMAX, WiDi or LTE.
Embodiment six:
The present embodiment is essentially identical with embodiment one, is particular in that:
With reference to Fig. 6.Described picture receiver is connected image decoder module (112) by wireless receiving end (111) and constitutes, and image decoder module (112) is used for decompressed image data, and view data is delivered to frame buffer (22).Whether its work simultaneously is controlled by microprocessor.Described wireless receiving end is linked in sequence is constituted by tuner (1111), low-noise amplifier (1112), frequency mixer (1113), radio-frequency tuner (1114), digital demodulation channel-decoding (1115) and demultiplexer (1116).Described picture decoding is linked in sequence by entropy code (1121), inverse quantizer (1122) and IDCT (1123), and has table specification (1124) to connect entropy code (1121) and table specification (1125) connection inverse quantizer (1122) composition.
Embodiment seven:
The present embodiment is essentially identical with embodiment one, is particular in that:
With reference to Fig. 1, Fig. 7 and Fig. 8.In above-mentioned intelligent wearable device based on parallel processor based on stochastic neural net, described frame buffer (22) is random access memory, and described frame buffer (22) can be SRAM SRAM or dynamic RAM DRAM.The view data that it decodes through image receiver module (1) for storage, and the view data transmission that can it be stored shows to micro-display (41).After the view data that image transmitting terminal sends over is stored by described frame buffer (22), when needing repeatedly to show this view data, image transmitting terminal need not send again, as long as directly reading from frame buffer (22).
Embodiment eight:
The present embodiment is essentially identical with embodiment one, is particular in that:
With reference to Fig. 7.In above-mentioned intelligent wearable device based on parallel processor based on stochastic neural net, described frame buffer (22) is made up of six modules, i.e. sram cell array (2201), line decoder (2203), column decoder (2205), sense amplifier (2206), control logic circuit (2207) and input buffer (2202).Sram cell array (2201) is the core of frame buffer (22), is used for storing data.Line decoder (2203) and column decoder (2205) produce row address and column address, and the intersection of row and column seeks to the memory element in the memory cell array (2201) accessed.Sense amplifier is used to, when reading the content of memory element, reduce the access time of memorizer, improves read or write speed and reduces power consumption.Control logic circuit produces control signal, controls the read-write operation of whole frame buffer (22).
Embodiment nine:
The present embodiment is essentially identical with embodiment one, is particular in that:
With reference to Fig. 8.In above-mentioned low-consumption wireless micro display system, described frame buffer (22) is made up of nine modules, i.e. input/output port (2219), write driver (2218), row input buffer (2212), column address decoder (2213), dynamic memory cell array (2211), row input buffer (2216), row refresh controller (2215), row-address decoder (2217), sense amplifier (2214).Input/output port (2219) is used for input and the output of data.Write driver (2218) is used for producing dynamic memory cell array (2211) and carries out writing the signal of data.Row input buffer (2212) deposits column address for temporarily, and column address decoder (2213) for being decoded drawing the column address of dynamic memory cell array (2211) to column address.Dynamic memory cell array (2211) is used for storing data.Row input buffer (2216) deposits row address for temporarily.Row refresh controller (2215) is for refreshing dynamic memory cell array (2211).Row-address decoder (2217) for being decoded drawing the row address of dynamic memory cell array (2211) to row address.The sense amplifier (2214) voltage variety when amplifying and storage unit carries out reading data.
Embodiment ten:
The present embodiment is essentially identical with embodiment one, is particular in that:
In above-mentioned intelligent wearable device based on parallel processor based on stochastic neural net, described micro-display (41) is a kind of miniature display device: silicon based LCD micro-display, silicon-based organic light-emitting micro-display, silicon-based inorganic luminescence micro-display, non-silicon-based luminescence micro-display or miniature CRT monitor, Diagonal Dimension is less than 3.3cm, pixel resolution is not less than 640*480, the gray scale of single color component is not less than 32 grades, support binocular 3D application, support brightness, contrast and color saturation adjust, support Gamma correction, support monochrome/YCbCr/RGB picture format, support PAL/NTSC/SMPTE video standard, support line by line or interlaced scan mode, support the input modes such as work VGA/DVI/RGB.It is received in control module (2) by the corresponding interface.If described micro-display (41) supports fractal scanning function, the most described micro-display (41) interface (F) answers integrated fractal scanning control module, in order to produce fractal scanning control signal, form more than 256 grades gray-scale Control signals and drive micro-display (41).
Embodiment 11:
The present embodiment is essentially identical with embodiment one, is particular in that:
With reference to Fig. 1.In above-mentioned intelligent wearable device based on parallel processor based on stochastic neural net, described can be photographic head (43), sound receiver (44), inertial sensor (45) etc., they access in corresponding interface (M), it is eventually connected to artificial neural network and microprocessor or microcontroller (21), for artificial neural network analysis identification, their work is controlled by microprocessor or microcontroller (21).Video transmission interface (G) can be VGA, DVI, RGB, HDMI, MIPI, MDDI, other kinds of LVDS, SPI, IIC, USB, other kinds of serial ports, parallel port etc..Audio transmission interfaces (H) can be analog audio interface such as TRS, RCA, XLR, it is also possible to be digital audio interface such as AES/EBU, S/PDIF, or USB interface.The interface (P) of inertial sensor (45) can be Special Interface Chip.
Claims (15)
1. an intelligent wearable device based on parallel processor, including image receiver module (1), control module (2), concurrent operation coprocessor (3), outside input-output equipment (4), external non-volatile storage device (5), it is characterized in that: described image receiver module (1) connects external non-volatile storage device (5) and control module (2), control module (2) connects concurrent operation coprocessor (3) and outside input-output equipment (4)；Described image receiver module (1) is for wirelessly receiving view data from image transmitting terminal (6)；Described control module (2) is made up of microprocessor or microcontroller (21), frame buffer (22), internal storage location (23) and each class interface, the control information that microprocessor or microcontroller (21) are used for producing image receiver module (1) and micro-display (41) needs, is the most also used for managing the work of concurrent operation coprocessor (3) and processing its data transmitted；Concurrent operation coprocessor (3) is for the Data Analysis Services transmitting external input device;Outside input-output equipment (4) includes micro-display (41) and photographic head (43), is used for showing image, obtaining external information；External non-volatile storage device (5) is used for the program that stores, data and parameter.
Intelligent wearable device based on parallel processor the most according to claim 1, it is characterized in that: described concurrent operation coprocessor (3) uses neural network structure, connect outside input-output equipment (4), microprocessor or microcontroller (21) and external non-volatile storage device (5).
Intelligent wearable device based on parallel processor the most according to claim 2, it is characterized in that: described neutral net includes input layer (A), intermediate layer (B) and output layer (C), wherein intermediate layer (B) is if being made up of dried layer, described input layer (A) is made up of I input neuron, and described input neuron (A) accepts the first data (AI), exports the first data sequence (AO)；(B) can have some sublayers in described intermediate layer, each sublayer is made up of Jn relay cell, described intermediate layer neuron (B) accepts the first data sequence (AO), argument sequence (R), complete RBF computing, export the second data sequence (BO)；Described output layer (C) is made up of K output neuron, described output neuron accepts the second data sequence (BO), argument sequence (R), exports the 3rd data sequence (CO), wherein I, Jn, K are the integer more than or equal to 1；The outfan of input neuron is connected to the input of relay cell, and the outfan of relay cell is connected to the input of output neuron, and it is connected by complete being connected or the most connected.
Intelligent wearable device based on parallel processor the most according to claim 2, it is characterized in that: described neural network structure is made up of stochastic neural net based on probabilistic operation, this neutral net by binary number by being converted to random data sequence, so that sequence occurring, the probability of 0 or 1 represents the numerical value used in deterministic parameters calculation, multiplying is completed with logical AND gate or logic XOR gate, exponent arithmetic is completed with one-dimensional linear state machine or two-dimensional matrix state machine, after network completes to calculate, then probit is converted to straight binary number；Described each data sequence is to occur in data sequence in a period of time that the probit of 0 or 1 represents；Each argument sequence is converted to random parameter sequence by random transition device (31)；Described relay cell is all with the data sequence of above one layer of output for inputting independent variable, with relevant parameter sequence as function parameter, complete RBF computing, calculating process all uses probability number, data sequence i.e. occurring in a period of time, the probability of 0 or 1 represents a numerical value, exports random data sequence.
5. according to based on parallel processor the intelligent wearable device described in claim 3 or 4, it is characterised in that: described RBF kind includes but not limited to Gaussian function, many quadratic functions, inverse many quadratic functions, thin plate spline function, cubic function, linear function.
Intelligent wearable device based on parallel processor the most according to claim 2, it is characterised in that: described neutral net includes but not limited to stochastic neural net, semi-random neutral net, BP neutral net, PNN neutral net, Hopfield network, SOM neutral net, LVQ neutral net.
Intelligent wearable device based on parallel processor the most according to claim 1, it is characterized in that: described microprocessor or microcontroller (21) embedded microprocessor, it is connected with image receiver module (1), internal storage location (23), concurrent operation coprocessor (3), frame buffer (22), internal storage location (23), external non-volatile storage device (5), outside input-output equipment (4), is used for producing the work of a series of control information management each equipment.
Intelligent wearable device based on parallel processor the most according to claim 1, it is characterized in that: described external non-volatile storage device (5) is FLASH storage chip, the interface (E) of this storage device is IDE or SATA interface, for storing all parameters required for artificial neural network, it is also used for the External memory equipment of microprocessor or microcontroller.
9. according to based on parallel processor the intelligent wearable device described in claim, it is characterised in that: described image receiver module (1) picture receiver (11)；Being connected image decoder module (111) by wireless receiving end (111) to constitute, image decoder module (111) is used for decompressed image data.
Intelligent wearable device based on parallel processor the most according to claim 1, it is characterized in that view data is sent to image receiver module (1) by described image transmitting terminal (6) by wireless communication protocol, the wireless communication protocol that its interface is supported can be UWB, WHDI, WirelessHD, WIFI, WiMAX, WiDi or LTE.
11. intelligent wearable devices based on parallel processor according to claim 1, it is characterised in that: described frame buffer (22) is random access memory, is SRAM SRAM or dynamic RAM DRAM.
12. intelligent wearable devices based on parallel processor according to claim 1, it is characterized in that: described micro-display (41) is a kind of miniature display device: silicon based LCD micro-display, silicon-based organic light-emitting micro-display, silicon-based inorganic luminescence micro-display, non-silicon-based luminescence micro-display or miniature CRT monitor, Diagonal Dimension is less than 3.3cm, pixel resolution is not less than 640*480, the gray scale of single color component is not less than 32 grades, support binocular 3D application, support brightness, contrast and color saturation adjust, support Gamma correction, support monochrome/YCbCr/RGB picture format, support PAL/NTSC/SMPTE video standard, support line by line or interlaced scan mode, support the input modes such as work VGA/DVI/RGB；It is received in control module (2) by the corresponding interface.
13. intelligent wearable devices based on parallel processor according to claim 1, it is characterized in that: described external input device (4), it is photographic head (43), access video transmission interface (G), being connected to microprocessor or microcontroller (21) and stochastic neural net, video transmission interface (G) can be VGA, DVI, RGB, HDMI, MIPI, MDDI, other kinds of LVDS, SPI, IIC, USB, other kinds of serial ports, parallel port etc..
14. intelligent wearable devices based on parallel processor according to claim 1, it is characterized in that: described external input device (4), it is sound receiver (44), access audio transmission interfaces (H), it is connected to microprocessor or microcontroller and stochastic neural net, audio transmission interfaces (H) is analog audio interface TRS or RCA or XLR, or digital audio interface AES/EBU or S/PDIF, or USB interface.
15. intelligent wearable devices based on parallel processor according to claim 1, it is characterized in that: described external input device (4), it is inertial sensor (45), for gathering this equipment moving status information, receive in the corresponding interface (P), being connected to microprocessor or microcontroller or stochastic neural net, the interface (P) of inertial sensor is Special Interface Chip.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610216172.1A CN105979201A (en) | 2016-04-11 | 2016-04-11 | Intelligent wearable device based on parallel processor |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610216172.1A CN105979201A (en) | 2016-04-11 | 2016-04-11 | Intelligent wearable device based on parallel processor |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105979201A true CN105979201A (en) | 2016-09-28 |
Family
ID=56989622
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610216172.1A Pending CN105979201A (en) | 2016-04-11 | 2016-04-11 | Intelligent wearable device based on parallel processor |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105979201A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018113216A1 (en) * | 2016-12-19 | 2018-06-28 | 威创集团股份有限公司 | Dlp system based on machine learning |
| CN110062279A (en) * | 2019-04-22 | 2019-07-26 | 努比亚技术有限公司 | Video method of cutting out, wearable device and computer readable storage medium |
| CN110135563A (en) * | 2019-05-13 | 2019-08-16 | 北京航空航天大学 | A kind of convolutional neural networks binarization method and computing circuit |
| CN110663048A (en) * | 2017-09-05 | 2020-01-07 | 松下电器（美国）知识产权公司 | Execution method, execution device, learning method, learning device, and program for deep neural network |
| CN111052134A (en) * | 2017-08-31 | 2020-04-21 | 快图有限公司 | Peripheral processing device |
| CN111507463A (en) * | 2019-01-30 | 2020-08-07 | 闪矽公司 | Neuromorphic symbol processor and method of operating the same |
| CN112633473A (en) * | 2020-12-18 | 2021-04-09 | 展讯通信（上海）有限公司 | Wearable device based on AI and application data processing method thereof |
| CN113727014A (en) * | 2020-05-25 | 2021-11-30 | 西克股份公司 | Camera and method for processing image data |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102657403A (en) * | 2012-05-28 | 2012-09-12 | 上海海事大学 | Electronic safety helmet and method for monitoring safety of location |
| CN102740076A (en) * | 2012-06-15 | 2012-10-17 | 上海大学 | Wireless micro-display system |
| US20140085501A1 (en) * | 2010-02-26 | 2014-03-27 | Bao Tran | Video processing systems and methods |
| CN104320615A (en) * | 2014-10-17 | 2015-01-28 | 智擎信息系统（上海）有限公司 | Intelligent video security and protection system and signal processing method thereof |
| CN104537684A (en) * | 2014-06-17 | 2015-04-22 | 浙江立元通信技术股份有限公司 | Real-time moving object extraction method in static scene |
-
2016
- 2016-04-11 CN CN201610216172.1A patent/CN105979201A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140085501A1 (en) * | 2010-02-26 | 2014-03-27 | Bao Tran | Video processing systems and methods |
| CN102657403A (en) * | 2012-05-28 | 2012-09-12 | 上海海事大学 | Electronic safety helmet and method for monitoring safety of location |
| CN102740076A (en) * | 2012-06-15 | 2012-10-17 | 上海大学 | Wireless micro-display system |
| CN104537684A (en) * | 2014-06-17 | 2015-04-22 | 浙江立元通信技术股份有限公司 | Real-time moving object extraction method in static scene |
| CN104320615A (en) * | 2014-10-17 | 2015-01-28 | 智擎信息系统（上海）有限公司 | Intelligent video security and protection system and signal processing method thereof |
Non-Patent Citations (2)
| Title |
|---|
| 张俊: "《信贷风险模型设计应用入门》", 31 January 2014 * |
| 曾黄麟: "《智能计算 关于粗集理论、模糊逻辑、神经网络的理论及其应用》", 30 June 2004 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018113216A1 (en) * | 2016-12-19 | 2018-06-28 | 威创集团股份有限公司 | Dlp system based on machine learning |
| CN111052134A (en) * | 2017-08-31 | 2020-04-21 | 快图有限公司 | Peripheral processing device |
| CN111052134B (en) * | 2017-08-31 | 2024-04-26 | 快图有限公司 | Peripheral processing apparatus |
| CN110663048A (en) * | 2017-09-05 | 2020-01-07 | 松下电器（美国）知识产权公司 | Execution method, execution device, learning method, learning device, and program for deep neural network |
| CN110663048B (en) * | 2017-09-05 | 2023-10-24 | 松下电器（美国）知识产权公司 | Execution method, execution device, learning method, learning device, and recording medium for deep neural network |
| CN111507463A (en) * | 2019-01-30 | 2020-08-07 | 闪矽公司 | Neuromorphic symbol processor and method of operating the same |
| CN110062279A (en) * | 2019-04-22 | 2019-07-26 | 努比亚技术有限公司 | Video method of cutting out, wearable device and computer readable storage medium |
| CN110135563A (en) * | 2019-05-13 | 2019-08-16 | 北京航空航天大学 | A kind of convolutional neural networks binarization method and computing circuit |
| CN113727014A (en) * | 2020-05-25 | 2021-11-30 | 西克股份公司 | Camera and method for processing image data |
| CN112633473A (en) * | 2020-12-18 | 2021-04-09 | 展讯通信（上海）有限公司 | Wearable device based on AI and application data processing method thereof |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105979201A (en) | Intelligent wearable device based on parallel processor | |
| CN104424652B (en) | Technology for reducing the access for retrieving texture image | |
| US7447997B2 (en) | Reducing information transfer in screen capture series | |
| CN101425277B (en) | Liquid crystal display, LCD driver, and operating method of lcd driver | |
| US11232767B2 (en) | Image display method, display system and computer-readable storage medium | |
| CN109522902A (en) | The extraction of Space-Time character representation | |
| CN101072352B (en) | Video data package method and system | |
| CN106817545B (en) | A kind of fast multiresolution video image mirror image rotation processing system | |
| US11183105B2 (en) | Display panel and device, image processing method and device, and virtual reality system | |
| CN106104668A (en) | Display device, the driver of display device, include display device and the electronic equipment of driver and display system | |
| US10452955B2 (en) | System and method for encoding data in an image/video recognition integrated circuit solution | |
| US20130120406A1 (en) | Graphics processing | |
| US10726815B2 (en) | Image processing apparatus, display panel and display apparatus | |
| CN104156191A (en) | System and method for extending mobile electronic device to PC display | |
| US20210279589A1 (en) | Electronic device and control method thereof | |
| CN101322393B (en) | Image processing device and method, image transmission device and method, and display device | |
| US20220408077A1 (en) | 3d display device, method and terminal | |
| US20190220699A1 (en) | System and method for encoding data in an image/video recognition integrated circuit solution | |
| US20060033753A1 (en) | Apparatuses and methods for incorporating an overlay within an image | |
| CN105451021A (en) | Low-power-consumption wireless micro-display system | |
| US20170193895A1 (en) | Low latency display system and method | |
| CN114217691B (en) | Display driving method and device, electronic equipment and intelligent display system | |
| US6661465B2 (en) | Television interface for handheld calculator for use with multiple calculator display formats | |
| WO2022042053A1 (en) | Data processing method and system, and electronic device | |
| Iglesias et al. | Enabling powerful GUIs in ISOBUS networks by transparent data compression |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20160928 |