Abstract: An image processing method includes the following steps. Firstly, a block-size value is obtained. Then, a first frame into a plurality of first blocks according to the block-size value is divided. Then, a second frame into a plurality of second blocks according to the block-size value is divided. Then, a noise reduction intensity array is obtained according to a first pixel information of each first block and a second pixel information of each second block. Then, an output frame is obtained according to the noise reduction intensity array, the first frame and the second frame.

Abstract: The present invention discloses an image processing method. The image processing method includes the following steps: (a), a to-be-processed image is corrected as a first correction image according to a first mapping relationship along a correction direction; (b) the first correction image by an angle is rotated; and (c) the rotated first correction image is corrected as a second correction image according to a second mapping relationship along the same correction direction. In embodiment, given that the to-be-processed image is deformed along two different directions, the to-be-processed image is corrected along the same correction direction, such that correction complexity could be reduced.

Abstract: An encryption and signature device for AI model protection is provided. The encryption and signature device for AI model protection includes a key derivation unit, a model encryption unit, a model password encryption unit, an image generation unit and a signature unit. The key derivation unit is configured to derive a model key according to a model password and a derivation function. The model encryption unit is configured to encrypt an AI model according to the model key to generate an encrypted AI model. The model password encryption unit is configured to encrypt the model password to generate an encrypted model password. The image generation unit is configured to generate an image file according to the encrypted model password and the encrypted AI model. The signature unit is configured to sign the image file according to a private key to obtain a signed image file.

Abstract: An image processing method includes: (a) receiving a to-be-processed image; (b) obtaining a target brightness of the to-be-processed image; (c) creating a histogram curve of the to-be-processed image; (d) clipping the histogram curve according to a clipping value; (e) obtaining a mapping function of the clipped histogram curve; (f) obtaining a mapped image mapped according to the mapping function; (g) obtaining an average brightness of the mapped image; (h) obtaining a difference between the target brightness and the average brightness; and (i) if the difference is greater than a threshold, adjusting the clipping value and repeating steps (d) to (i) until the difference is less than the threshold. Thus, if the to-be-processed image does not reach the target brightness, the image processing device automatically repeats steps (c) to (i) until the to-be-processed image reaches the target brightness.

Abstract: An image processing method includes the following steps. Firstly, a block-size value is obtained. Then, a first frame into a plurality of first blocks according to the block-size value is divided. Then, a second frame into a plurality of second blocks according to the block-size value is divided. Then, a noise reduction intensity array is obtained according to a first pixel information of each first block and a second pixel information of each second block. Then, an output frame is obtained according to the noise reduction intensity array, the first frame and the second frame.

Abstract: The present invention discloses an image processing method. The image processing method includes the following steps: (a), a to-be-processed image is corrected as a first correction image according to a first mapping relationship along a correction direction; (b) the first correction image by an angle is rotated; and (c) the rotated first correction image is corrected as a second correction image according to a second mapping relationship along the same correction direction. In embodiment, given that the to-be-processed image is deformed along two different directions, the to-be-processed image is corrected along the same correction direction, such that correction complexity could be reduced.

Abstract: A mixed-precision artificial intelligence (AI) processor and an operating method thereof are provided. The AI processor includes a first calculation module, a second calculation module and a control module. The first calculation module is configured to perform calculation based on the data with a first format. The second calculation module is configured to perform calculation based on the data with a second format different from the first format. The control module is coupled to the first calculation module and the second calculation module to select one of the first calculation module or the second calculation module to perform calculation based on an input data according to a calculation strategy.

Abstract: An image processing method includes: (a) receiving a to-be-processed image; (b) obtaining a target brightness of the to-be-processed image; (c) creating a histogram curve of the to-be-processed image; (d) clipping the histogram curve according to a clipping value; (e) obtaining a mapping function of the clipped histogram curve; (f) obtaining a mapped image mapped according to the mapping function; (g) obtaining an average brightness of the mapped image; (h) obtaining a difference between the target brightness and the average brightness; and (i) if the difference is greater than a threshold, adjusting the clipping value and repeating steps (d) to (i) until the difference is less than the threshold. Thus, if the to-be-processed image does not reach the target brightness, the image processing device automatically repeats steps (c) to (i) until the to-be-processed image reaches the target brightness.

Abstract: An encryption and signature device for AI model protection is provided. The encryption and signature device for AI model protection includes a key derivation unit, a model encryption unit, a model password encryption unit, an image generation unit and a signature unit. The key derivation unit is configured to derive a model key according to a model password and a derivation function. The model encryption unit is configured to encrypt an AI model according to the model key to generate an encrypted AI model. The model password encryption unit is configured to encrypt the model password to generate an encrypted model password. The image generation unit is configured to generate an image file according to the encrypted model password and the encrypted AI model. The signature unit is configured to sign the image file according to a private key to obtain a signed image file.

Abstract: A computing device for floating-point mathematic operation using look-up table is provided. The computing device includes: a bit arrangement unit used for receiving a floating-point input data and performing a bit arrangement or a format conversion on the floating-point input data to generate multiple index blocks; a first look-up table unit group used for receiving the index blocks and performing look-up operation using the index blocks as index to generate a plurality of look-up table results; and an operation unit used for performing operation on the look-up table results of the first look-up table unit group to generate an operation output.

Abstract: A data sharing method that implements data tag to improve data sharing on a multi-computing-unit platform, wherein the multi-computing unit platform includes multiple cores, and multiple threads generating multiple critical sections on each core. When a first thread enters a first critical section to access a shared data, the shared data is temporarily stored in a first core, when the first thread leaves the first critical section, it transfers the control of the shared data to a second core that has higher transmission advantage.

Abstract: The present invention discloses a method of using memory allocation to address hot and cold data, which comprises steps: using a hardware performance monitor (HPM) to detect at least one read/write event of a central processor; while a number of the read/write events reaches a threshold or a random value, a computer system recording an access type of the read/write event occurring latest and a memory address causing the read/write event; and the computer system assigning the memory object in the memory address to a volatile memory or a non-volatile memory according to the memory address and the access type. Thereby, data pages can be assigned automatically according to the access types categorized by the central processor, exempted from being assigned manually by engineers.

Abstract: The present invention discloses a method of using memory allocation to address hot and cold data, which comprises steps: using a hardware performance monitor (HPM) to detect at least one read/write event of a central processor; while a number of the read/write events reaches a threshold or a random value, a computer system recording an access type of the read/write event occurring latest and a memory address causing the read/write event; and the computer system assigning the memory object in the memory address to a volatile memory or a non-volatile memory according to the memory address and the access type. Thereby, data pages can be assigned automatically according to the access types categorized by the central processor, exempted from being assigned manually by engineers.

Abstract: A distorted image correcting method is applied to a distorted image correcting apparatus to correct a distorted image to further generate a corrected image. The image correcting apparatus includes a buffer, which includes a memory block having a capacity that is smaller than a size of a data block of the distorted image. The distorted image correcting method includes: controlling a memory controller to retrieve M sets of segment data of the data block to the memory block, where M is smaller than N; and generating a part of the corrected image according to the M sets of segment data.

Abstract: A distorted image correcting apparatus corrects a distorted image to generate a corrected image. The distorted image correcting apparatus includes: a buffer, including a plurality of memory blocks each storing one of a plurality of blocks of data of the distorted image; an allocation circuit, allocating the memory blocks in the buffer according to a memory configuration, wherein the memory blocks have different memory capacities; a memory controller, retrieving the blocks of data of the distorted image from a memory and storing the blocks of data to the buffer according to the memory configuration; and a correction circuit, retrieving a block of data from a memory block of the buffer, and generating a part of the corrected image according to the block of data.

Abstract: Provided is a method for fast booting/shutting down a computing system. The method includes the steps of: when the computing system enters the hibernation mode, sorting the memory of the computing system into swappable pages and non-swappable pages and writing the non-swappable pages into a hibernation file of a storage device; determining whether the swappable pages are frequently-used pages or infrequently-used pages, and if the swappable pages are determined to be frequently-used pages, incorporating the frequently-used pages in the hibernation file; sorting the infrequently-used pages into clean pages and dirty pages; capturing one of the dirty pages and adding pages that are related to the captured page into at least one data set and placing the data set into a swap space of the storage device by continuous accessing process.

Abstract: A method for initializing a peripheral device and an electronic device using the method. The electronic device includes one or more peripheral devices having registers, a memory having a data storing module, and an instruction capturing module. The instruction capturing module captures a plurality of hardware register settings from a driver execution process of the one or more peripheral devices, stores the plurality of hardware register settings in the data storing module, and serializes or concatenates the plurality of hardware register settings to form serialized hardware register settings, when the electronic device is performing a non-hibernation resume or non-wakeup cold boot to execute an initialization process of the one or more peripheral devices. The one or more peripheral devices are initialized by the serialized hardware register settings, when the electronic device is performing cold boot again due to a hibernation resume or wakeup to execute the initialization process.

Abstract: A fast speed computer system power-on & power-off method, that is used to reduce an amount of main memory transferred and stored from a main memory into a second storage device, thus speeding up a speed of re-activation of a computer system from a hibernation state into a full speed operation state. Said fast speed computer system power-on & power-off method is applicable to various types of computer systems, and can be used to write in and load back data in cooperation with a random access processing technology. In addition, said method can be used to reduce extent of data loss and damage of said computer system due to a sudden power outage of said computer system.

Abstract: A multi-core electronic system for accessing a data storage device includes a plurality of processors, a data transmission interface and a rate adjustment module. The processors respectively provide a bandwidth requirement, and communicate with the data storage device via the shared data transmission interface. The rate adjustment module receives the bandwidth requirements, and determines a transmission rate of the data transmission interface according to the bandwidth requirements.

Abstract: A Hadamard transform-based image compression method includes performing a Hadamard transform on 2k pixel values according to a product of a 2k×2k stage Hadamard matrix and a 2k×2k adjustment matrix to generate 2k conversion values, where k is a positive integer and at least one of the 2k conversion values is zero. The adjustment matrix satisfies a condition that: when the 2k pixel values are divided into G pixel groups each comprising 2k/G adjacent pixels values and the adjustment matrix is multiplied with a first 2k×1 matrix formed by the 2k pixel values to transform the first 2k×1 matrix to a second 2k×1 matrix, each pixel value of the first matrix is transformed to an average of a pixel group comprising the pixel value to form the second 2k×1 matrix.