8results about How to "Understand technical means" patented technology

The present disclosure provides a picture processing method and device, electronic equipment and storage medium, wherein the method comprises: obtaining a source picture with source pixel values and a target picture without target pixel values; splitting the target picture based on a preset split size to obtain a plurality of target sub-pictures; determining a source sub-picture corresponding to the target sub-picture in the source picture based on the target sub-picture, and judging whether the pixel amount contained in the source sub-picture is consistent with a preset threshold; if yes, determining each target pixel value in the target sub-picture based on each source pixel value in the source sub-picture. The present disclosure splits based on the preset split size and then performs picture processing, so that the pixel amount contained in the source sub-picture corresponding to the split target sub-picture can better meet the preset threshold set for SARM, thereby ensuring the conversion relationship between the source picture and the target picture in the case of effective splitting, and the overall picture processing performance is better.

This disclosure provides an entity classification method, apparatus, electronic device, and storage medium. The method includes: acquiring multiple entity relationship networks corresponding to multiple entity nodes; the same entity node corresponding to different entity relationships in different entity relationship networks; performing multi-label prediction on the multiple entity relationship networks based on multiple trained base learners to obtain the multi-label prediction result of each entity node in each entity relationship network; and determining the final multi-label prediction result corresponding to each entity node based on the multi-label prediction result of each entity node in each entity relationship network. This disclosure, from the perspective of multi-view learning, performs multi-label prediction based on multiple entity relationship networks, which can more fully explore the relationships between entities, making the multi-label prediction results for entity nodes more accurate.

This disclosure provides a microphone array noise reduction method, apparatus, system, electronic device, and storage medium. The method includes: acquiring an audio signal to be denoised; performing frequency domain processing on the audio signal to obtain a frequency domain signal; responding to a first sub-frequency domain signal with a frequency less than a preset threshold in the frequency domain signal, performing a first noise reduction process based on a pre-set subtractive microphone array to obtain a denoised first sub-frequency domain signal; responding to a second sub-frequency domain signal with a frequency greater than or equal to a preset threshold in the frequency domain signal, performing a second noise reduction process based on a pre-set additive microphone array to obtain a denoised second sub-frequency domain signal; and obtaining a denoised audio signal based on the denoised first and second sub-frequency domain signals. This disclosure combines the superior characteristics of subtractive and additive microphone arrays to perform targeted noise reduction processing on low-frequency and high-frequency signals respectively, resulting in better noise reduction performance.

The present invention provides a virtualization system, method and device calling method for a trusted execution environment, the system comprising: a driver virtual host, running in a trusted execution environment TEE, for reusing a shared device driver developed by a general driver framework within the trusted execution environment TEE, so as to provide a shared driver service in the trusted execution environment TEE; at least one driver semi-virtualization module, running in a TEE virtual machine in a rich execution environment REE and / or an executable environment TEE, for calling a shared driver service provided by the driver virtual host, and providing a corresponding device driver service in the rich execution environment REE and / or the TEE virtual machine based on the shared driver service. By using the above system and method, it is possible to avoid repeated development of related device drivers, reduce the complexity of driver development, and reduce the cost of driver development.

The utility model provides a hydrogen turbocharging system for a fuel cell, which comprises an air compressor, a hydrogen storage device, a hydrogen turbine and a fuel cell stack, an output end of the air compressor is connected to the fuel cell stack, an output end of the hydrogen storage device is connected to an input end of the hydrogen turbine, and an output end of the hydrogen turbine is connected to an output end of the fuel cell stack. The output end of the hydrogen turbine is connected to the fuel cell stack, the hydrogen turbine is arranged on the air compressor and rigidly connected to a crankshaft of the air compressor, and the hydrogen storage device is used for providing high-pressure hydrogen for the hydrogen turbine. An air compressor is used to receive air and compress air with electrical energy and mechanical energy from a hydrogen turbine driven by high pressure hydrogen, and the high pressure hydrogen passing through the hydrogen turbine forms decompressed hydrogen, the compressed air and the decompressed hydrogen being provided to a fuel cell stack. According to the scheme, electric power for driving the air compressor can be saved, the size of the system is reduced, and the manufacturing and maintenance cost of the system is reduced.

The present invention provides a signal orientation method, device, system and computer-readable storage medium, the method comprising: using a sensor array to obtain a multi-channel signal to be oriented; for each preset direction in a plurality of preset directions, performing the following steps: determining the first signal feature of the multi-channel signal based on each preset direction, determining the second signal feature of the multi-channel signal based on the opposite direction of each preset direction, and determining the difference value between the first signal feature and the second signal feature; comparing the difference values ​​corresponding to the plurality of preset directions, and determining the preset direction corresponding to the maximum difference value as the target signal direction. Using the above method, the orientation error caused by the sensor response distortion can be reduced, with good adaptive anti-noise capability, and more accurate signal orientation can be achieved.

The application provides a manufacturing method of a PCIE golden finger and an electroplating lead structure. The method comprises: arranging the golden finger on a PCB board, wherein the golden finger comprises a ground network golden finger and a non-ground network golden finger; connecting the ground network golden finger with a ground hole located on a first side of the ground network golden finger; arranging a first electroplating bus on the PCB board, wherein the first electroplating bus transversely connects all the ground holes on the first side of the golden finger together and extends out of a second side edge and / or a third side edge of the PCB board; arranging a first electroplating branch on the PCB board, wherein the first electroplating branch connects the non-ground network golden finger in the golden finger to the first electroplating bus; electroplating the golden finger by using the first electroplating bus and the first electroplating branch; and drilling a hole on the first electroplating branch to cut off the connection between the non-ground network golden finger and the first electroplating bus after the electroplating is completed. The manufacturing process of the golden finger is simple and the residual lead on the PCB board is less.