17 results about "Dct transform" patented technology

The invention discloses a radar data compression and transmission method based on wavelet lifting-DCT (Discrete Cosine Transformation)-mixed entropy coding. The method comprises the following steps: processing data into a two-dimensional matrix format; the network condition is monitored in real time, and the network is selected according to the availability and bandwidth characteristics of the network; setting a dynamic compression ratio according to a network environment; performing biorthogonal wavelet transformation on the two-dimensional data in the row direction, performing biorthogonal wavelet transformation on the transformed result in the column direction, and removing redundant information; a block-shaped DCT method is adopted, DCT is further conducted on the low-frequency part and the high-frequency part after wavelet transformation, coefficients are effectively concentrated to the upper left corner of a matrix, energy distribution is concentrated, the compression efficiency is improved, and meanwhile the block size is dynamically determined by monitoring the network state according to the parameters and a dynamic compression ratio setting rule; a mixed entropy coding method is adopted, and lossless compression is carried out by using data statistical characteristics; and transmitting the compressed data.

A method of encoding a video stream in a video encoder is provided that includes computing an offset into a transform matrix based on a transform block size, wherein a size of the transform matrix is larger than the transform block size, and wherein the transform matrix is one selected from a group consisting of a DCT transform matrix and an IDCT transform matrix, and transforming a residual block to generate a DCT coefficient block, wherein the offset is used to select elements of rows and columns of a DCT submatrix of the transform block size from the transform matrix.

A solid-state image data storage method compresses image data when writing and decompresses it when reading. The image data compression process involves inputting image data, segmenting the image, performing a fast DCT transform, performing threshold quantization, and obtaining compressed data. The image data decompression process involves inputting compressed data, segmenting the data, performing a fast inverse DCT transform, and obtaining a decompressed image. This method compresses image data at the master control level, enabling high-speed storage. This method maximizes the amount of image data written into limited NAND storage, extending the NAND's lifespan.

The invention relates to the technical field of image coding, in particular to a variable-granularity self-adaptive coding method for a smoke and dust disaster environment, which comprises the following steps: performing dust infrared transmissivity dynamic compensation based on the obtained current dust concentration and visibility of a smoke and dust disaster environment scene, and calculating a transmissivity compensation coefficient; synchronously acquiring a visible light image and an SWIR image of the smoke and dust disaster environment scene, and performing gain compensation on the SWIR image based on the transmissivity compensation coefficient to obtain a compensated SWIR image; performing feature extraction and fusion on the visible light image and the compensated SWIR image to obtain fusion features, and restoring the fusion features into a fusion image; selecting the size of a DCT block in combination with the local texture complexity, and performing DCT on the fused image to obtain a group of DCT coefficients; and applying quantization parameter threshold constraint to the abnormal high-frequency DCT coefficient, and carrying out variable granularity DCT quantization on the fused image to obtain entropy coding.

The present invention discloses a medical image tampering detection and self-recovery method based on texture cross-embedding. The method divides a medical image into a ROI region and a RONI region. Texture complexity is calculated based on different features in the ROI region, which is then divided into texture blocks and smooth blocks. Compressed sensing technology is used to extract different features within each block as recovery information. Pixel-level and block-level detection bits are set in the ROI region. Restoration information is hidden in the RONI region using a reference matrix and cross-embedding technology. This method not only uses gradient histograms and DCT transforms to detect copy-and-paste attacks across the entire image, but also uses pixel-level and block-level detection to locate tampered blocks in the ROI region and extract recovery information to restore the tampered image. The method offers the advantages of strong tamper detection, excellent self-recovery, high security, good invisibility, and strong robustness.

A method of encoding a video stream in a video encoder is provided that includes computing an offset into a transform matrix based on a transform block size, wherein a size of the transform matrix is larger than the transform block size, and wherein the transform matrix is one selected from a group consisting of a DCT transform matrix and an IDCT transform matrix, and transforming a residual block to generate a DCT coefficient block, wherein the offset is used to select elements of rows and columns of a DCT submatrix of the transform block size from the transform matrix.

This application provides a method, apparatus, and device for image dark watermarking based on DCT intermediate frequency difference modulation and pseudo-random spreading. The method performs Hamming error correction coding and key-driven pseudo-random spreading processing on the watermark information to be embedded to generate a spread-spectrum watermark bit sequence; it performs channel extraction and frequency domain representation generation processing based on discrete cosine transform on the original image to obtain a DCT transform coefficient matrix; based on the spread-spectrum watermark bit sequence, it performs intermediate frequency coefficient pair difference modulation embedding processing on the DCT transform coefficient matrix to generate a modulated frequency domain coefficient matrix carrying watermark feature information; it performs image domain inverse transformation reconstruction on each modulated frequency domain coefficient matrix and outputs a target image containing a dark watermark that is visually consistent with the original image. This application, through intermediate frequency difference modulation and spreading mechanisms, solves the shortcomings of traditional schemes in terms of visual invisibility, resistance to compression attacks, and security, significantly improving the robustness of the watermark.

This invention discloses a digital image watermark embedding and extraction method, device, and storage medium. The method combines histogram shifting with neural network embedding. After the DCT transform, histogram shifting is performed to minimize noise in the embedded data stream under JPEG compression. A neural network compensation algorithm is then used to minimize the impact of the data stream on the host image. This technical solution addresses the problem of poor performance of existing neural network digital watermarking techniques against JPEG compression attacks.

This invention discloses a high-capacity robust video watermarking method resistant to geometric attacks, comprising: acquiring the YUV three channels of a video; applying DTCWT transform to the U channel to separate the coefficients in each sub-band; selecting a watermark embedding strategy, dividing the coefficients into blocks and applying DCT transform, modifying the DCT coefficients according to the watermark information to be embedded, and sequentially applying inverse DCT transform and inverse DTCWT transform to obtain the watermarked U channel; combining it with the Y and V channels of the original video and performing the inverse YUV conversion operation, finally obtaining the watermarked video through video encoding. The extraction process is based on DCT transform, and the watermark is extracted according to the embedding strategy. This invention achieves robust watermark embedding and extraction based on DTCWT and DCT, flexibly balancing the robustness of the watermark against geometric attacks with the capacity of the payload, and can be applied to multimedia platforms where video is the mainstream.

A method of encoding a video stream in a video encoder is provided that includes computing an offset into a transform matrix based on a transform block size, wherein a size of the transform matrix is larger than the transform block size, and wherein the transform matrix is one selected from a group consisting of a DCT transform matrix and an IDCT transform matrix, and transforming a residual block to generate a DCT coefficient block, wherein the offset is used to select elements of rows and columns of a DCT submatrix of the transform block size from the transform matrix.

The present invention belongs to the field of office automation technology and discloses an email security detection method and system based on frequency domain and semantic analysis, an electronic device, and a computer-readable storage medium. The method comprises obtaining an email to be detected and preprocessing images in its attachments; performing a block operation on the preprocessed image, then performing a DCT transform on each block to generate a corresponding DCT coefficient matrix; calculating the high-frequency component entropy value #imgabs0# corresponding to each block based on the DCT coefficient matrix; determining the steganographic suspicion detection result of the email based on the #imgabs1# value; triggering SMTP traffic semantic analysis of the email based on the steganographic suspicion detection result to obtain a comprehensive risk score #imgabs2#; and triggering a multi-level alarm linkage mechanism based on the comprehensive risk score #imgabs3#. The present invention achieves a high detection rate, low latency, and highly adaptable data theft protection for emails, providing reliable technical support for enterprise-level network security.

The invention belongs to the technical field of artificial intelligence, and discloses a neural network watermark embedding and extracting method and device, equipment and a storage medium. The embedding method comprises the following steps: acquiring an original watermark and a secret key; randomly selecting a plurality of weight parameters from the full connection layer weight of the neural network model to be embedded with the watermark according to the secret key; performing DCT (Discrete Cosine Transform) on the selected weight parameters to obtain a DCT coefficient of each weight parameter; embedding a watermark into the selected DCT coefficient; generating a digital signature for the original watermark, and embedding the original watermark and the digital signature into a neural network model; and performing inverse DCT conversion on the DCT coefficient after the watermark is embedded into a weight parameter after the watermark is embedded, and loading the weight parameter after the watermark is embedded into a neural network model. According to the method, the DCT technology, the QIM technology and the ECDSA technology are organically combined to achieve a synergistic effect, and key challenges in an existing neural network watermarking technology can be effectively solved, including the problems of influence on model performance, insufficient robustness and safety.

The present invention discloses a medical image lossless compression method and system based on transform coding and deep learning. The method includes the following steps: performing a DCT transform and round quantization on an input 2D medical image to split it into DC coefficients and AC coefficients; compressing the loss R caused by the DC coefficients and round quantization; decomposing the AC coefficients into structure layer MSBs and texture layer LSBs; decomposing the texture layer LSBs into multiple sub-coefficient blocks; constructing a deep probability estimation model based on a convolutional neural network, performing group-based autoregressive probability density modeling, inputting multiple sub-coefficient blocks, and outputting a probability distribution matrix for each sub-coefficient block; and performing lossless encoding and decoding on each sub-coefficient block by combining an entropy arithmetic encoder and the probability distribution matrix. The present invention has the advantages of high compression efficiency and lossless compression.

This application relates to the field of data compression technology, and in particular to a method, apparatus, device, and storage medium for seismic data compression. The method includes: acquiring co-offset seismic data; performing a two-dimensional DCT transform on the co-offset seismic data to obtain frequency-wavenumber domain data; processing the frequency-wavenumber domain data based on the SPECK compression algorithm to obtain primary coded data; and performing adaptive arithmetic coding on the primary coded data to obtain target compressed data. This application facilitates improved efficiency in real-time transmission and storage of seismic data.

The application discloses a device for realizing fast two-dimensional DCT transformation and a method thereof. The device comprises a bit line driving unit, a word line driving unit, a source line driving unit, a photoelectric storage and calculation integrated device array unit, an integration unit, a current-voltage conversion and analog-digital conversion unit and a shift accumulation unit. The driving units are used for inputting a transformation matrix and a data vector. The photoelectric storage and calculation integrated device array unit is used for completing multiplication operation of the transformation matrix and the data vector. The integration unit is used for integrating positive value components and negative value components into one data. The current-voltage conversion and analog-digital conversion unit is used for converting a photoelectric current signal into a voltage signal and converting a voltage analog signal into a digital signal. The shift accumulation unit is used for shifting and accumulating intermediate results. The device can realize fast two-dimensional DCT transformation, and when the hardware is used to realize calculation acceleration, the area, power consumption and calculation delay of the hardware architecture are greatly reduced.

The invention discloses an image steganography method and device based on a pseudo-random generator. The method comprises the following steps: acquiring a carrier image, converting the carrier image into a YCrCb color space and extracting a brightness channel to obtain a brightness matrix, and performing 8 * 8 partitioning on the brightness matrix to generate a pixel block matrix; generating a pseudo-random sequence based on the date vector and the task number through a pseudo-random generator, and determining an embedded block sequence list; embedded blocks in the pixel block matrix are determined according to the embedded block sequence list, 2D-DCT conversion and quantization are carried out on the embedded blocks, then Z-shaped scanning is carried out, a 1 * N frequency domain coefficient vector matrix is obtained, the first vector of the frequency domain coefficient vector matrix is a DC coefficient, and the rest are AC coefficients; sequentially splicing the data to be steganographic into an embedded information bit string, and embedding the embedded information bit string into the AC coefficient of the frequency domain coefficient vector; and performing inverse Z-shaped scanning, inverse quantization and 2D-IDCT transformation on the frequency domain coefficient vector matrix of the embedded data to reconstruct a secret-carrying image. According to the scheme, the security of image steganography can be improved.