75 results about "Channel gating" patented technology

The invention relates to the technical field of exercise rehabilitation, and discloses a percutaneous spinal nerve electrical stimulation system and method, and the system comprises a power module which is used for converting an external alternating current into a stabilized power supply; the MCU module controls the pulse signal generation module; the pulse signal generation module is used for generating an electrical stimulation waveform; the DAC module receives the digital signal; the modulation module is used for receiving analog signals; the amplification module is used for receiving modulation waveforms; the multi-channel gating module is used for distributing stimulation current to a plurality of channels; the current detection module monitors the magnitude of current output by each channel in real time; and the ADC module performs digital conversion on the current analog signal output by the electrical stimulator. According to the invention, multi-channel and real-time modulation electrical stimulation signals can be output through modulation of two paths of stimulation signals, spinal cord sensory nerves are activated through the non-invasive epidermal electrode, amputation hand feeling is induced, the defect that electromyographic signals are interfered by the tail end of a stump in the prior art is improved, and the accuracy of an electromyographic signal decoding algorithm is also improved.

The invention relates to the technical field of communication, in particular to a USB-A interface and serial port dynamic switching and protection system based on an analog switch. Comprising a main control unit, the main control unit is used for planning dynamic switching and conflict coordination of a USB-A interface and a serial port as a whole, generating a conflict-free channel switching control instruction by integrating a dynamic priority scheduling algorithm and combining a USB data transmission real-time load, a serial port communication request emergency degree and a device historical interaction record fed back by a state detection unit, and sending the conflict-free channel switching control instruction to the state detection unit; the analog switch unit is driven to complete channel gating; an analog switch unit; a grading protection unit; a protocol adaptation unit; and a state detection unit. Priority scheduling is carried out by means of the main control unit in combination with multi-dimensional parameters such as the transmission load of the USB-A interface and the communication emergency degree of the serial port UART, communication channels of the USB-A interface and the serial port can be flexibly switched, the flexibility and adaptability of interface switching are enhanced, and the problem of switching limitation caused by a single triggering condition is solved.

The invention relates to an improved YOLOv8 small target detection method based on SENet, and the method comprises the steps: introducing semantic dilution loss, and measuring the dilution degree of small target features in a channel; a suppression reverse weight is generated through an SENet structure, and background redundancy is suppressed while a high response area is reserved; a C2f structure of YOLOv8 is fused, and decoupling characteristics are transmitted in cross-layer connection; in the Neck feature pyramid, a SENet response migration relation is constructed; calculating a weight deviation value, and judging whether the small target response has spatial deviation or not; position balance loss is introduced, and feature repositioning is carried out on a small target area with overlarge center-of-gravity drift; designing a channel response consistency measurement index, and measuring the SE response consistency of the small target between different epochs; sENet channel output is extracted from the image, and the variance of target area channel response distribution is counted; a channel with high variability is weakened or suppressed from a current image detection path, a detection head is introduced into a channel gating mechanism, a stable channel is adaptively selected to participate in prediction, and the small target feature representation capability is significantly enhanced.

The invention discloses a wearable optical sensor glove, a system, a gesture recognition method and an article recognition method. Comprising a customized flexible glove body, a flexible circuit board, a photoelectric detector, a flexible force-induced luminescence optical fiber, a multipath gating device and a single-chip microcomputer control module. Wherein the customized flexible glove body is made of a flexible material; the flexible circuit boards are arranged on finger parts and a hand back part of the glove; the photoelectric detector is mounted in the finger region and is electrically connected with the multi-path gating device; the flexible force-induced luminescence optical fiber is fixed on the photoelectric detector in each finger region; the multi-channel gating device and the single-chip microcomputer control module are also integrated on the flexible circuit board on the back of the hand and are electrically communicated with each other. The system has good anti-interference performance, and can realize the functions of high-speed dynamic detection, article identification and the like.

The invention relates to a method for monitoring prefabrication and assembly errors of a steel box laminated continuous box girder bridge, and belongs to the technical field of data processing. The method comprises the following steps: collecting prefabrication and assembly error monitoring data of the steel box superposed continuous box girder bridge, and marking the data; an error prediction model is constructed, wherein the error prediction model comprises a physical constraint feature fusion layer, a causal convolution layer, a Bayesian uncertainty propagation layer, a residual attention mechanism module and a dual-channel gating fusion module; training the model by adopting the marked data; in the training process, optimizing the model by constructing mixed loss to obtain a trained model; and aligning timestamps of the collected structure parameters, construction processes, environment parameters and measurement and state parameters of the new prefabricated segment, integrating the parameters into an input sample isomorphic with the training set, and inputting the input sample into the trained model to obtain a prediction result. According to the invention, systematic deviation can be effectively eliminated, and high-robustness and high-precision error prediction is realized.

A chess-like piece positioning and recognition system, including: an MCU, an RFID module, multi-channel analog switches, an electromagnet and communication signal switch unit, an electromagnetic drive unit, antennas, silicon steel blocks, and pieces. The MCU is used for communication with the RFID module, selecting channels of the multi-channel analog switches, and electromagnet and communication signal switch selection. The RFID module is used for reading information from the pieces. The multi-channel analog switches are used for group gating and intra-group channel gating of the RF channels. The electromagnet and communication signal switch unit is used for switching between the electromagnetic drive unit and the RF channels. The electromagnetic drive unit is used for driving an electromagnet. The antenna is used for sending and receiving signals from the pieces and also serves as the coil for the electromagnet. The silicon steel block is used as the core of the electromagnet, which generates a magnetic attraction to the pieces after being magnetized.

The invention belongs to the technical field of anesthesia depth regulation and control, and particularly relates to an anesthesia depth self-adaptive regulation and control method and system based on a dynamic time sequence model, and the system comprises a portable computer which is used for operating a program, receiving electroencephalogram time sequence data output by a signal collection device, receiving a pressing event of a handle, and obtaining efficacy information of an effect part; a channel gating result is generated based on the information and used for conducting category division on a drug administration adjustment request, gating is conducted through a shared code LSTM and a two-channel LSTM head and an effect room concentration vector and a hysteresis band, the problem of rhythm dislocation caused by exponential smoothing lag and induction-waking asymmetry is solved, mistaken up-regulation or mistaken down-regulation and overshoot are reduced, and the accuracy of drug administration adjustment is improved. And the adjustment convergence time is shortened, so that the anesthesia depth is more stably maintained in a target interval.

The invention belongs to the field of discrete device testing, and particularly discloses a discrete device testing system resource self-adaptive extension device and method. According to the application, the number of channel gating is set according to the tested device, so that the test system resource is matched with the circuit of the tested device; the gating state of the test channel and the resource switching sequence are judged by controlling the switching unit, and the switching unit is linked with the test system to automatically switch the channel, so that the time division multiplexing of the resources of the discrete device test system is realized; and then the test system resources are connected to the pins of the tested device in a customized manner through the resource expansion interface, so that the resource expansion of the discrete device test system can be realized. The method has a good application prospect under the problems of insufficient discrete device test system resources and low test efficiency. Experiments do not need to frequently replace testing system programs and clamps, and testing system resources are effectively expanded.

The invention discloses a voltage inspection system of a fuel cell or an electrolytic tank and a method thereof. The voltage inspection system comprises a plurality of acquisition modules, a high common mode differential operational amplifier module, a signal conditioning module, an MCU module and an isolation power supply module, each acquisition module comprises a module selection circuit and a multi-path check circuit and is used for acquiring data after channel gating; the input end of the high common mode differential operational amplifier module is electrically connected with the module output end of each acquisition module, and the high common mode differential operational amplifier module is used for subtracting the potentials on the gating lines at the two ends of the acquired channel to obtain a gating potential difference value; the input end of the signal conditioning module is electrically connected with the output end of the high common-mode differential operational amplifier module, and the signal conditioning module is used for conditioning the gating potential difference value transmitted by the high common-mode differential operational amplifier module to obtain conditioning data; the input end of the MCU module is electrically connected with the output end of the signal conditioning module, and the MCU module is used for collecting conditioning data of the signal conditioning module and transmitting the conditioning data to an upper computer through an isolated CAN communication network.

The invention discloses a double-flow fault diagnosis method based on state space modeling and dynamic edge context awareness, the method constructs a time domain state flow and a time frequency edge flow, and the specific implementation process is as follows: the time domain state flow is subjected to state space modeling under a Transform framework to form a MambaT encoder, long and short term dependence is efficiently captured with linear complexity, and the time domain state flow is subjected to dynamic edge context awareness; time domain sensitive features are accurately represented; the time-frequency edge stream converts vibration data into a two-dimensional time-frequency image through wavelet transformation, a differentiable edge sensitive mask is designed to adaptively separate time-frequency image impact characteristics and background noise, boundary content perception adaptive filling is designed to maintain image boundary topology integrity, and the two parts cooperate to form dynamic edge context perception convolution, so that the image boundary topology integrity is improved. Time-frequency edge features are effectively extracted; and a channel enhancement and channel gating module is designed to establish a dynamic channel interaction gating mechanism, so that time domain-time frequency domain double-current cascade characterization is realized, and the fault discrimination capability is remarkably improved. Experiments prove that the method has higher fault diagnosis performance.

The invention discloses a board card test system and a board card test method, and relates to the technical field of power electronics. In the board card test system, an upper computer is used for sending a corresponding target excitation signal to a to-be-tested signal board card based on an HIL equipment signal list document; sending a channel gating instruction to the fault injection board card based on the current test channel information of the to-be-tested signal board card; sending a test switching instruction to a signal test module based on a target to-be-tested signal type corresponding to the target excitation signal; the to-be-tested signal board card is used for outputting a corresponding target response signal in response to the target excitation signal; the fault injection board card is used for conducting a target channel in response to the channel gating instruction; the signal test module is used for accessing a target test instrument corresponding to the target to-be-tested signal type in response to the test switching instruction; and transmitting a test result of the target response signal to the upper computer. According to the embodiment of the invention, the automatic excitation test of the HIL board card can be more efficiently and accurately realized.

The application relates to the technical field of communication, in particular to a USB-A interface and serial port dynamic switching and protection system based on an analog switch. The system comprises a master control unit, an analog switch unit, a hierarchical protection unit, a protocol adaptation unit and a state detection unit. The master control unit is used for overall planning of dynamic switching and conflict coordination of the USB-A interface and the serial port, generates conflict-free channel switching control instructions by integrating a dynamic priority scheduling algorithm, combining USB data transmission real-time load, serial communication request urgency and device historical interaction records fed back by the state detection unit, and drives the analog switch unit to complete channel gating. The application relies on the master control unit, carries out priority scheduling in combination with multi-dimensional parameters such as transmission load of the USB-A interface and communication urgency of the serial port UART, can flexibly switch the communication channels of the USB-A interface and the serial port, enhances the flexibility and adaptability of interface switching, and solves the switching limitation problem caused by a single triggering condition.

The invention belongs to the technical field of equipment control, and relates to a wormwood essential oil extraction equipment control method and system. According to the method, the distillation state is comprehensively sensed by collecting the temperatures of multiple positions in a kettle, the heating steam pressure, the liquid phase temperature of an outlet of a condenser and an acoustic emission signal in the distillation process in real time; then inputting the signal into a two-channel gating circulation network model which is dynamically modulated by track entropy and can be fused with a physical model to carry out Bayesian correction under a stable working condition, so as to accurately measure and obtain a soft measurement value of the oil yield of the wormwood essential oil; then, based on the oil yield deviation, a multi-objective optimization function containing the square of the deviation, the control quantity change rate, the thermal shock gradient and the boiling instability risk at the same time is solved, and a control sequence is obtained; and finally, decoupling the control sequence, and determining and executing an accurate distribution strategy for the pressure and flow of the heating steam. On the premise of ensuring stability and safety, the extraction efficiency of the wormwood essential oil is maximized.

The invention belongs to the technical field of biological feature recognition, and discloses an automatic recognition and automatic switching system based on a multifunctional USB peripheral interface, which comprises a function module, a state detection module, a USB signal switching module and a main control module, the function module is used for accessing an external device and feeding back a voltage signal to the main control module and the state detection module; the state detection module is used for judging whether external equipment is accessed or not and judging the type of the accessed external equipment; the USB signal switching module is used for opening an internet access channel or a PC / OTG channel; the beneficial effects of the invention are as follows: 1, hardware-level channel gating is realized, software intervention is not needed, and rapid switching is realized; 2, a special detection chip is not needed, and the cost is reduced; and 3, the interface resource utilization rate is improved, and the system redundancy is reduced. And 4, three scenes of the internet access, the PC and the OTG are automatically identified, user configuration is not needed, and the compatibility is good.

The invention discloses a hardware acceleration deployment method of an ultrasonic full-focus algorithm on an FPGA (Field Programmable Gate Array) platform, and relates to the technical field of deployment of an acoustic imaging algorithm on the FPGA platform, a master-slave structure taking the FPGA platform as a master controller is adopted, and a plurality of FPGA slaves are controlled to collect and calculate parameters in parallel through channel gating; a full-focus parameter calculation process is modularly divided, the calculation throughput rate is improved through a pipeline method, and calculation results of the multiple FPGA slaves are transmitted to the FPGA host for imaging; furthermore, a mirror image full-focus calculation method is adopted, parameter multiplexing is achieved through a recombination calculation process, and the parameter calculation time is shortened by half. The problems that an ultrasonic full-focusing imaging algorithm is complex in calculation, long in time consumption and difficult to deploy in a lower computer platform can be solved, development of a low-cost and light-weight ultrasonic phased array full-focusing online detection instrument is facilitated, and the method has important significance on maintenance and health monitoring of a pressure container.

The invention relates to a small target detection method and system based on adaptive texture perception and frequency domain-cross-layer collaborative optimization, and the method comprises the steps: providing a cross-scale texture guide module, independently capturing the spatial distribution characteristics of heterogeneous features through a multi-scale texture extraction unit, and constructing scene-level texture prior through an additive fusion mechanism; a frequency domain detail enhancement module is provided, features are mapped to a complex frequency spectrum space through real number fast Fourier transform, dynamic reweighting is carried out on medium-high frequency components by using an adaptive weight vector, and target contour information is explicitly enhanced in combination with inverse transform and a residual connection mechanism; a cross-layer feature stabilization module is designed, context clues of different granularities are captured through lightweight multi-scale receptive field extension, and a dynamic channel gating modulation mechanism is introduced to perform active intervention on a feature evolution path. Compared with the prior art, the method has the advantages of characteristic enhancement, stable positioning, balanced detection precision and operation efficiency and the like.

The invention discloses a dual-band broadband direction finding system and method based on antenna group gating. The dual-band broadband direction finding system comprises an antenna array, a signal conversion unit, a channel gating unit, a digital processing unit and a control and interaction unit, the antenna array is used for integrally covering the first antenna group and the second antenna group based on unified geometric framework arrangement, and realizing unified output of cross-band direction finding results; the signal conversion unit is used for converting a radio frequency signal into a signal which can be processed by the digital processing unit; the channel gating unit is used for realizing configurable connection between the first antenna group or the second antenna group and a receiving channel of the digital processing unit; the digital processing unit is used for completing signal sampling, feature extraction and direction estimation operation and outputting a direction finding result; and the control and interaction unit is used for outputting parameters and results. According to the invention, the high-frequency direction-finding antenna group and the low-frequency direction-finding antenna group are respectively constructed under a unified hexagonal geometric framework, so that the system gives consideration to omnidirectional coverage, direction-finding sensitivity and engineering implementation feasibility under the condition of a wide frequency band.

The embodiment of the application provides a kind of multi-channel data acquisition method and system, comprising: receiving the multi-channel digital signal after analog-digital conversion unit conversion;Wherein, multi-channel digital signal is to the analog signal input according to the preset strobe mode Channel selection, after the analog signal selected is converted into corresponding digital signal by analog-digital conversion unit, it is obtained;Interpolation filter is used to carry out interpolation filter processing to multi-channel digital signal, and the digital signal after multi-channel synchronization is obtained;Wherein, the interpolation filter is constructed according to the number of input channel and the number of channel sampling data points, and the number of channel sampling data points is the number of data points collected by analog-digital conversion unit for each switching of channel.The application can solve the problem of different synchronization of multi-channel data acquisition based on multiplexing structure.

The utility model discloses an integrated laser drive circuit for an OTDR optical module and the optical module. The integrated laser drive circuit comprises a laser chip, a laser drive chip, a pulse generation circuit and a microcontroller. The laser driving chip comprises n paths of high-speed communication driving channels and n paths of OTDR driving channels which are correspondingly connected with the n paths of lasers in the laser chip respectively so as to realize time division multiplexing of the n paths of lasers. And the microcontroller controls the laser driving chip to gate and switch between the high-speed communication driving channel and the OTDR driving channel through a register arranged in the laser driving chip, and controls the pulse generation circuit to transmit a pulse signal for OTDR testing to the laser driving chip. According to the optical module, a high-speed switching chip does not need to be used, switching between a high-speed communication function and a network diagnosis function is achieved through cooperation of the microcontroller and the built-in register of the laser driving chip, the circuit integration degree is high, and miniaturization packaging of the optical module is facilitated.

The invention belongs to the technical field of radar radiation source identification. The invention provides a radar radiation source identification method based on relation perception and prototype optimization. According to the embodiment of the invention, through channel gating and a soft attention mechanism, invalid frequency band features are automatically inhibited, and the weight of a bad sample which deviates from the center is reduced. And a GCN is introduced to construct a sparse prototype graph, so that the model can perceive a similarity relationship between categories, and local consistency is enhanced through a graph propagation mechanism. And in cooperation with prototype comparison loss, centers of similar categories are forcibly pushed away, so that a clearer isolation strip is formed at a decision boundary, and the false identification rate is remarkably reduced. A residual mean fusion structure is also designed. When an attention mechanism may fail due to extremely small sample size, a mean value prototype on a residual path can play a role in guaranteeing the bottom, so that the stability of model training is ensured; and when the sample size is slightly large, the attention mechanism can provide finer feature expression, so that the universality of the model under different small sample settings is realized.

The application discloses a crop nitrogen content detector and relates to the technical field of spectrum analysis. A reference signal conditioning circuit in the detector provides a reference signal for a phase-locked demodulation circuit, a multi-channel gating circuit sequentially guides constant-current drive signals to a plurality of LEDs according to interval time set by a microcontroller, and drives the plurality of LEDs to alternately emit AC light signals carrying DC signals to a leaf surface; a photoelectric processing circuit sequentially receives reflected light signals of different wave bands from the leaf surface, filters the DC signals in the reflected light signals, and transmits the retained reflected AC signals to the phase-locked demodulation circuit; the phase-locked demodulation circuit demodulates the retained reflected AC signals of different wave bands according to the reference signal, and outputs DC signals that are linearly proportional to the amplitude of the AC light signals in the active light signals; and the microcontroller obtains the spectral reflectivity of the leaf surface to each wave band according to the DC signals of different wave bands. The application can greatly improve the efficiency of measuring the nitrogen content of crops.

The invention discloses an acupoint positioning method and system based on image enhancement, and belongs to the technical field of image processing, and the method comprises the steps of human body acupoint image collection, human body acupoint image enhancement, acupoint recognition model construction and acupoint positioning. According to the scheme, through parameter optimization of the gamma value and the fractional order, according to the difference between the oscillation coefficient and the individual position, random disturbance is introduced for updating, the parameter combination is continuously optimized in combination with the evolution rate, and gamma transformation and fractional order enhancement are performed by using the optimal parameter combination, so that the image quality is improved; the activation and recombination unit performs intelligent screening on features through channel gating activation and then performs splitting and recombination on a feature map to obtain a composite feature map, and the bitype convolution optimization unit obtains a core convolution feature map and a detail convolution feature map by splitting the composite feature map and combining group convolution with point-by-point convolution. And weighting is carried out through global average pooling and a soft attention mechanism, so that the accuracy and the stability of acupoint recognition are remarkably improved.

The invention provides a cross-border fund flow risk assessment method which can be applied to the technical field of artificial intelligence. The method comprises the following steps: acquiring a standardized event stream; for the standardized events in the time window, constructing an incremental graph structure used for representing fund link flow characteristics; in the incremental graph structure, node fission strength is calculated for an initial edge in the current time window, a node with the node fission strength exceeding a first preset threshold value serves as a fission starting point to execute three-hop breadth-first search, and a search result is obtained; performing dual-channel gating convolution processing on the transaction sequence and the regional feature sequence represented by the candidate sub-graph determined based on the target path to obtain a sub-graph embedded vector; and splicing the sub-graph embedding vector with the predicted merging residual duration, the node fission strength and the cross-regional hop count into a state vector, inputting the state vector into a game reinforcement learning model to output the freezing probability of a candidate sub-graph, and determining a risk assessment result of the standardized event according to the freezing probability.

The application discloses a communication device of optional communication channel, comprising one or more UART receiving ports, one or more UART sending ports, a port gating module; the port gating module comprises an input end and an output end, wherein the number of the input end of the port gating module is the same as the number of the UART receiving ports, and the number of the output end of the port gating module is the same as the number of the UART sending ports; the UART receiving ports are connected with the input end of the port gating module in one-to-one correspondence, and the UART sending port module is connected with the output end of the port gating module in one-to-one correspondence; the port gating module is used for connecting one or more target input ends with one or more target output ends in response to a channel gating instruction. The configurable gating, switching and expansion among multiple channels are realized, and the flexible and complex bus data transmission requirement is met.

The application discloses an atomizing device, which comprises a shell assembly, a first atomizer, a second atomizer and a gating structure. The shell assembly is surrounded to form a mounting cavity and is provided with an air inlet channel communicated with the mounting cavity. The first atomizer, the second atomizer and the gating structure are arranged in the mounting cavity. The first atomizer comprises a first atomizing assembly, and the first atomizing assembly is provided with a first atomizing channel. The second atomizer comprises a second atomizing assembly, and the second atomizing assembly is provided with a second atomizing channel. The gating structure is used for selectively connecting one of the first atomizing channel and the second atomizing channel with the air inlet channel in fluid communication. The atomizing power of the first atomizing assembly is greater than that of the second atomizing assembly. After the first atomizing assembly is connected with the air inlet channel in fluid communication through the gating structure, the second atomizing assembly can generate a second aerosol with a small atomizing amount at a time. The second aerosol is a smokeless aerosol, has very low visibility or is almost invisible, and can avoid affecting others in public places as much as possible.

The application relates to the field of artificial intelligence and feature enhancement technology, and provides a multi-head dynamic soft gating method based on cross-layer visual attention propagation trajectory bias, which comprises the following steps: acquiring a visual token sequence; calculating the weight of multi-head attention and single-head output; constructing a trajectory vector containing at least attention intensity, active attention intensity, entropy, change rate, divergence, historical gating, CLS bidirectional attention, CLS rollout accumulation and spatial neighborhood mean by using cross-layer attention; layer-in standardization; projecting into a bias injection current token representation; combining the bias representation and the trajectory to generate a layer-by-layer, head-by-head, token-by-token and channel-by-channel gating; writing into a cache, and obtaining an effective gating through strength coefficient conversion, and then modulating channel by channel before splicing; splicing and modulating output and projecting. The method can inhibit redundant propagation, retain token sequences and enhance foreground representation.

The application discloses a kind of double-channel rolling bearing fault diagnosis methods of image time sequence fusion, belong to rotating machinery intelligent fault diagnosis technical field.It includes: collecting rolling bearing vibration signal, constructs fixed length one-dimensional time sequence sample by segmented pretreatment, and generates corresponding image by gram angle difference field transformation;Then respectively establish image channel and time sequence channel, image channel extracts overall spatial feature and local detail feature, time sequence channel uses LSTM to extract time sequence dependent feature, and sets fusion node in multilevel;Further, the corresponding level features of two channels are complementarily fused based on joint context channel-by-channel gating, obtain joint fault feature layer by layer, finally input classification module to complete fault identification.The application solves the problem of insufficient single feature expression and insufficient cross-modal information fusion in rolling bearing fault diagnosis, and improves the diagnosis accuracy and robustness.

The utility model relates to a resource board card and a test machine, and the resource board card comprises a main board card which is connected with a communication board card, receives a plurality of clock signals with different frequencies outputted by the communication board card, and transmits the clock signals to a sub-board card; the at least one sub board card comprises a multi-channel gating device and a control module, the multi-channel gating device is connected with the main board card and the control module, and the received clock signals with different frequencies are gated and transmitted to the control module; the clock signal is used for the control module to test the device to be tested. A plurality of clock signals with different frequencies are transmitted to the daughter board card through the main board card, the clock signals with the corresponding frequencies can be selected in the daughter board card through the multi-channel gating device according to test requirements, the control module uses the selected clock signals to test the device to be tested, clock signals do not need to be generated one by one for serial test, and the test efficiency is improved.

The invention discloses a cryoablation needle target area frosting state identification method based on image processing, and relates to the technical field of cryoablation needles. Constructing a target area original image sequence under the unified space-time coordinate system; outputting corresponding degradation stage information; obtaining a high-fidelity target region clear image sequence; aggregating the signal subspace features through a channel gating mechanism to obtain a signal enhancement feature image set; generating a target area multi-scale semantic feature tensor; outputting an initial frosting probability distribution diagram of the target area; performing form constraint segmentation optimization to obtain a frosted area mask with a consistent form; and a time sequence stable frosting area mask sequence is obtained, a frosting state dynamic evolution model is constructed, and early warning information is displayed in real time. According to the method, the problem of feature submerging caused by mixed noise is effectively inhibited, and the recognizable degree and stability of bottom layer structure information of a target area are remarkably improved in a low-temperature imaging environment.

A method for detecting pedestrians by improving RT-DETR, comprising the following steps: step 1: selecting a public pedestrian detection dataset Widerperson, then cleaning and retaining suitable parts; step 2: image enhancement and preprocessing are performed on the image data, and the training set, the validation set and the test set are divided according to the preset proportion; step 3: an enhanced real-time target detection model iRT-DETR is constructed, the model comprising a stripe convolution channel gating module (SCCG) in a backbone network, an adaptive scale decomposition attention module (ASDA) in an encoder, and a bidirectional multi-assistant feature pyramid network (BiMA-FPN); step 4: the training set obtained in step 2 is input into the model constructed in step 3 for training, and the final training convergent detection model is obtained through loss calculation, back propagation and hyperparameter adjustment; step 5: the trained model is used to detect input pedestrian images, and the class and boundary box position of the pedestrian target are output.