90 results about "JPEG" patented technology

The invention discloses a railway driver driving behavior monitoring method based on artificial intelligence. An artificial intelligence processing unit completes hardware interface initialization; the integrated wireless equipment of the railway locomotive sends a network IP address and equipment identification information of the main control end to a host process of the artificial intelligence processing unit; a host process of the artificial intelligence processing unit initiates a video stream pulling request to a digital camera at a corresponding end through an RTSP protocol; the main process of the artificial intelligence processing unit receives the RTSP code stream and then decodes the RTSP code stream into a picture in a JPEG format, the image is cut and zoomed, and complete metadata information is added for each picture; three reasoning processes of the artificial intelligence processing unit run in parallel, and each process performs targeted reasoning detection on the decoded JPEG picture and metadata based on a self-loaded model: a main process receives a reasoning detection result, performs comprehensive judgment in combination with a preset threshold value and an abnormal state duration, and sends the result to the main process; triple alarms of local reminding, driver informing and remote monitoring are realized, and the driving behavior of the driver is standardized.

A kind of railway driver driving behavior monitoring method based on artificial intelligence, artificial intelligence processing unit completes hardware interface initialization;Railway locomotive comprehensive wireless device sends the network IP address of main control end, device identification information to the main process of artificial intelligence processing unit;Artificial intelligence processing unit main process initiates video stream pull request to the digital camera of corresponding end by RTSP protocol;Artificial intelligence processing unit main process decodes into JPEG format picture after receiving RTSP code stream, trims zoom image, adds complete metadata information to each picture;Three inference processes of artificial intelligence processing unit run in parallel, each process is based on the model loaded by itself to decode the JPEG picture and metadata for targeted inference detection:Main process receives inference detection result, carries out comprehensive judgment in conjunction with preset threshold and abnormal state duration, realizes local reminder, driver notification, remote monitoring triple alarm, standardizes driver driving behavior.

The application provides a JPEG data decoding method and device, equipment and storage medium, which can be used in the field of image processing. The method comprises the following steps: obtaining target JPEG data, marking the target JPEG data based on a preset segmentation length in units of target MCUs to obtain target JPEG segmented data; storing the target JPEG segmented data into a target SVM to obtain multiple initial decoded segments and synchronization parameter information; in response to the multiple initial decoded segments being correctly decoded, judging whether the number of target MCUs corresponding to the multiple initial decoded segments meets a preset number condition; if yes, sending a target decoding instruction to a target GPU to enable the target GPU to perform target Huffman decoding to obtain multiple target decoded segments; decoding each target DCU included in each target decoded segment to obtain target JPEG decoding data; and storing the target JPEG decoding data into the target SVM to obtain a target RGB image corresponding to the target JPEG decoding data based on the target GPU. The CPU running load is reduced, and the decoding efficiency is improved.

An image steganography analysis method based on a hybrid deep learning framework combines the feature extraction advantage of a JPEG steganography rich model and the integratable advantage of a deep learning model to construct a multi-stage hybrid deep learning JPEG steganography analysis framework, and the framework has expandability and can be applied to image steganography analysis. The latest deep learning steganalysis network can be added to the framework. Meanwhile, the steganalysis problem under various complex scenes such as data source mismatching can also be adapted; a color image steganalysis CNN trunk model formed by three stages of preprocessing, feature extraction and feature classification is designed, and a space attention mechanism and a channel attention mechanism are added in a trunk stream, so that the detection accuracy is improved; a new algorithm optimization module is designed based on the framework, the network structure is adjusted, and the generalization ability, convergence performance, detection accuracy and robustness of the model are further improved.

The invention provides a residual DCT (Discrete Cosine Transformation) feature guided artifact learning module and an image tampering detection system, and the module comprises a DCT coefficient processing unit, a quantization table processing unit, a residual DCT calculation unit, a feature splicing unit and a neural convolution processing unit. The residual DCT calculation unit is used for performing repeated recompression on DCT coefficient matrix simulation JPEG compression to obtain DCT residual information and construct a residual DCT feature map on the basis of the DCT residual information, and the DCT coefficient processing unit is used for encoding the DCT coefficient matrix into DCT plane binary volume representation by using threshold truncation and One-Hot encoding, and then performing feature extraction and processing; the quantization table processing unit is used for expanding and processing a quantization table; the feature splicing unit is used for feature splicing; and the neural convolution processing unit is used for processing the spliced features to obtain final DCT features. According to the method, the positioning and identification capability of the tampered area in the double JPEG image with the same QF compressed twice is effectively improved.

The invention relates to a non-standard JPEG (Joint Photographic Experts Group) coding processing method for dynamic code rate control. The method comprises the following steps: presetting a multi-frame joint image processing scene, obtaining an original reference image, and carrying out combined processing of boundary expansion and random bias addition to obtain a preprocessed reference image; dividing macro blocks for the preprocessed reference image, and distributing an independent coding parameter configuration space after verification and compliance; frequency domain conversion is executed according to a division result, an initial quantization parameter is calculated through an index model, and a dynamic quantization table is dynamically adjusted; executing non-standard entropy coding to obtain an indexed macro block coding code stream; combining the indexed code stream and a macro block division result to construct an index table containing a macro block serial number, an initial address and a code stream length; analyzing the code stream to obtain quantized data, and performing inverse quantization by using a dynamic quantization table to obtain spatial domain data; enabling the format of the reconstructed image to be consistent with that of the preprocessed reference image through reverse preprocessing, and performing real-time processing on the linkage input image to obtain a reconstructed reference image; and non-standard JPEG coding processing of dynamic code rate control is realized.

A system and method for energy-efficient approximate digital JPEG and MJPEG-compression. The system includes a controller unit to control a processing loop for processing image blocks based on a comparison of a current image block to a previous image block. The system includes a quantization unit configured to quantize the frequency domain representation using an approximate quantization process and a quantization (Q) matrix. The quantization unit is configured to: identify, a nearest power of two value for each element of the quantization matrix; generate an updated Q matrix by assigning each element of the quantization matrix with the identified nearest power of two value; and shift each element of the updated Q matrix by a number of bits to generate a quantized frequency domain representation. The number of bits corresponds to the identified nearest power of two for the corresponding element of the updated Q matrix.

The application discloses a method for urine test paper mobile phone image detection and analysis, which comprises the following steps: collecting urine test paper JPEG images under multiple light sources under a standard light source box, taking the test paper image under the D65 light source as a standard image, converging the RGB color performance of the test paper block in the corresponding area of the test paper image under the standard light source to the test paper image under the remaining light source environment, that is, training and obtaining a compensation model, fully considering the test paper color features and the mobile phone camera characteristic features in the model characteristic parameters, ensuring the accuracy of the color correction model by analyzing the maximum value, minimum value and variance of the features during preprocessing, establishing a training TabNet model, fusing multiple regression models such as LightGBM, ElasticNet Regression and Support Vector Regression to optimize the color correction effect, and realizing a strengthened compensation model. Through the color correction of the test paper image by the compensation model, the error caused by the color performance of the urine test paper image affected by the environmental light is reduced, the most real color of the test paper image is restored, the color recognition degree is improved, the accuracy of the test paper detection and analysis is improved, and the model is small and has a fast recognition speed.

The present application belongs to the field of image processing and target detection, and specifically provides a RAW image processing method for target detection, which is used to improve the compression efficiency of the image, and can significantly improve the recognition accuracy when the processed image is applied to the target detection task. The present application constructs a semantic fusion network to directly generate an image from the RAW image, which is more suitable for detection and occupies less storage. In the present application, the RAW image retains a large amount of original information, and the semantic information is obtained through the JPEG image and the label provided thereby, which is integrated into the RAW image processed by the ISP. While taking into account the detection accuracy, the required number of bits for image transmission is effectively reduced. At the same time, an efficient and close-to-training-target loss function is used to constrain the entire training process, which can significantly reduce the space occupied by the image and improve the detection rate. In summary, after introducing the RAW image processing method for target detection, the present application can improve the recognition accuracy and reduce the demand for storage space.

The invention provides a real-time three-dimensional reconstruction method and system based on unmanned equipment. In order to solve the lag problem that in traditional reconstruction, collection needs to be conducted firstly and then reconstruction needs to be conducted, and the composite bottleneck that airborne computing power is limited and weak network bandwidth is insufficient, a unified time signal is generated through a hardware synchronization module, and microsecond-level synchronization of multi-source data is achieved; aiming at heterogeneous data characteristics, the airborne end respectively constructs batch processing assembly lines of point cloud octree compression, image JPEG coding and IMU byte alignment, so that the data volume is remarkably compressed while the computing power overhead is reduced; a QoS mechanism and local persistent cache of an MQTT protocol are introduced into a transmission layer, and reliable arrival of data during network fluctuation is guaranteed through a breakpoint resume technology. In conclusion, end-to-end real-time processing of software and hardware collaboration is realized, bandwidth occupation is effectively reduced, and stable and high-precision real-time three-dimensional reconstruction is realized in a weak network environment.

The application relates to a kind of implementation methods of FPGA-based JPEG-LS encoder, sequentially including context modeling, edge detection, adaptive prediction correction, prediction residual calculation, Golomb coding parameter K calculation, prediction error mapping, Golomb coding is carried out to prediction error MErrval, context parameter update and code stream splicing.This application can realize JPEG-LS lossless high compression rate encoder function on the basis of hardware without the aid of software.One aspect, discard the run length coding in JPEG-LS standard, reduce the complexity of hardware implementation, reduce the consumption of the hardware resource of the designed encoder, the application also obtains the encoder with compression rate of 18%~25% by optimizing the calculation of golomb coding parameter K and golomb coding mode.

The invention provides an infrared image watermark adding method and system, and the method comprises the following steps: obtaining an original infrared image, and carrying out the analysis of the original infrared image, so as to obtain the infrared radiation intensity of the original infrared image; generating a watermarking-free base map according to the infrared radiation intensity; moving the watermarking-free base map to an image display area, and performing zooming processing on the watermarking-free base map in the image display area, so that the watermarking-free base map and the image display area are consistent in size; and drawing a watermark on the watermark-free base map in the image display area to generate an infrared image with the watermark. According to the method, a JPEG or JPG image is not directly used as a watermark to draw a background image, the YUV image is zoomed, and the watermark is finally added, so that the definition of the watermark can be ensured, the infrared image after the watermark is drawn can still have high resolution, and the problem that the watermark drawing display on the infrared low-resolution image is blurred is solved.

The application discloses a JPEG image multi-carrier steganography method based on a monkey king evolution algorithm, which is used for embedding secret information into images and comprises the following steps: processing N images according to a user input password and an error correction matrix H based on a (7, 4) Hamming code to obtain N carrier images, wherein a matrix of the i-th carrier image is H i ′, i = 1, 2,..., N; converting secret information to be embedded Secret into a binary form, and encrypting to obtain a binary sequence M = (m1, m2,..., m L L is the length of the binary sequence M; taking the minimum difference between a carrier image feature set and a stego image feature set as an objective, using a monkey king evolution algorithm to perform non-uniform dynamic allocation on the load size of each carrier image, and according to the allocation result, cutting and writing the secret information into the corresponding carrier image to obtain a stego image set. The application can solve the problem of efficient allocation of secret information among multiple carrier images, so that the stego image is least detected.

This invention discloses a satellite image compression method under extremely low bandwidth conditions. The steps are as follows: collect a static image dataset; establish a variable bitrate image encoding / decoding network; train the encoding / decoding network; embed a generative adversarial network (GAN) at the decoder and retrain it; embed a super-resolution network after the encoder / decoder and retrain it; the user selects a region of interest (ROI), and the selected region is compressed; the decoder restores the bitstream to an image and displays it; the unselected regions are compressed; when bandwidth is sufficient, the unselected regions are transmitted, and the decoder stitches the images together; super-resolution processing is performed. This image transmission method, under extremely low bandwidth constraints, provides a superior subjective reconstruction effect compared to traditional image coding methods such as JPEG and deep learning-based image coding methods in a large number of test images.

The application discloses a lossy compression detection method and terminal fusing high-frequency space domain and DCT domain, and the method comprises the steps of obtaining a first JPEG compressed image data set, training a WebP lossy compression detection network according to the first JPEG compressed image data set to obtain a trained WebP lossy compression detection network, obtaining a second JPEG compressed image data set, performing WebP lossy compression trace detection on the second JPEG compressed image data set according to the trained WebP lossy compression detection network, and outputting a detection result; the method for extracting features of a JPEG image by fusing high-frequency space domain and DCT domain provided by the application solves the problem that the effect of the lossy compression detection method for JPEG double compression in the prior art is poor and even fails in some cases.

The embodiment of the invention discloses a bill analysis method and system. The method comprises the following steps: receiving an external request; analyzing a head feature code of the bill data, if a JPEG or PNG file identifier is detected, routing the bill data to an image processing channel, calling an OCR engine to perform full-graph identification on the bill data, and based on a province template corresponding to the provincial administrative region code, extracting a preset field from the bill data as an analysis result; if the PDF identifier is identified, routing the bill data to a PDF processing channel, and executing text layer extraction and seal detection on the bill data in parallel to obtain an analysis result; and according to the province template, mapping an analysis result into a standard JSON structure, performing field integrity, logic rationality and cross-field relevance verification on the standard JSON structure, and outputting the standard JSON structure after the verification is passed. According to the embodiment of the invention, the recognition precision of the bill data can be remarkably improved, the processing efficiency of bill analysis is comprehensively optimized, and the resource utilization efficiency is improved.

The invention belongs to the technical field of data processing, and discloses a color JPEG reversible watermark hiding method and system based on adaptive STC, and the method comprises the steps: obtaining a color JPEG image, decoding the color JPEG image, converting the color JPEG image into a Y color space, and extracting the DCT coefficients of the Y channel and the Y channel; according to the statistical characteristics of the DCT coefficients of all the channels, the embedding adaptability weight is calculated, and the embedding capacity is adaptively allocated to the three channels in combination with a preset channel weighting adjustment factor; a candidate set is selected according to the sequence of the three channels and the distributed embedding capacity, and zero value coefficients in the candidate set form a carrier sequence; the watermark to be hidden is embedded into the carrier sequence through STC coding, the secret-carrying JPEG image is generated, and high-capacity, low-distortion and completely reversible watermark hiding facing the color JPEG image is achieved.

The invention discloses a high-throughput JPEG (Joint Photographic Experts Group) heterogeneous reasoning and hybrid parallel method on mobile equipment, and provides an FCG reasoning framework aiming at the problems of high decoding time consumption proportion, high cross-processor communication overhead and low processor resource utilization rate in the existing mobile terminal JPEG image recognition. And optimization is carried out by combining JPEG frequency domain information with characteristics of a heterogeneous processor of a mobile terminal. The framework comprises an off-line preprocessing stage, an off-line task division stage and an on-line task scheduling stage, wherein in the off-line preprocessing stage, processor operation data are collected, and a delay prediction model is established; in the off-line task division stage, a processor cluster is constructed, model tasks are distributed, and the task amount is adjusted; and in the online task scheduling stage, hybrid parallel pipeline identification is executed according to the scheme. According to the method, through multi-core Huffman decoding, heterogeneous network structure design and optimization of task scheduling, the identification delay is remarkably reduced, the throughput is improved, the energy consumption is reduced, and the batch JPEG image identification performance of the mobile terminal is effectively improved.

The present application relates to the technical field of image processing, and particularly relates to an image processing method, an electronic device and a readable storage medium, the image processing method comprises the following steps: fusing and blocking a current frame and a reference frame to obtain a plurality of to-be-encoded spatial image blocks, and combining the to-be-encoded spatial image blocks into a plurality of macroblocks, and configuring a corresponding encoding quantization table and a decoding quantization table for each macroblock; performing discrete cosine transform by using a first integer DCT transform kernel; performing quantization by using the encoding quantization table; performing entropy encoding and storing into an off-chip memory; reading a compressed code stream from the off-chip memory, performing entropy decoding, and performing inverse quantization by using the decoding quantization table; performing inverse discrete cosine transform by using a second integer DCT transform kernel; performing block recombination to obtain a reconstructed reference frame, and taking the reconstructed reference frame as a new reference frame, taking a next frame input as a new current frame, and recycling. The present application can effectively solve the technical problem of distortion accumulation in the multi-round iteration encoding and decoding process based on JPEG.

Various systems and methods are presented regarding generating executable computer code / instructions from input files, whereby the input files may be an image file (e.g., JPEG, PDF, etc.). The image file can be digital capture of a sequence of instructions such as a graphical representation comprising a ladder diagram, a function block diagram, a sequential function chart, etc. P&ID and suchlike can also be submitted to the system. Code generated from the input files can be enhanced by application of historical data comprising pertinent subroutines, and suchlike. Further, an entity can be prompted to provide further information in the event of the input file does not provide all of the content.

The present application relates to a kind of phytoplankton imaging flow trait analysis method and system, wherein the method comprises: the single particle image exported by imaging flow cytometry equipment is defined as a phenotype object to be analyzed;From the JPEG annotation section of single particle image, corresponding optical signal field of phenotype object is obtained by byte analysis;Phytoplankton target region segmentation is carried out on single particle image, and phytoplankton target mask is obtained;The two-dimensional morphological characteristics of phenotype object are calculated based on phytoplankton target mask;Two-dimensional morphological characteristics and optical signal field are bound in corresponding phenotype object, and event-level morphological-optical fusion trait vector of phenotype object is generated, and event-level morphological-optical fusion trait vector is output.Using the present application, image morphological information and optical property information are synchronously acquired and fused from the same data source, and the acquisition efficiency of phytoplankton single particle comprehensive characteristics is improved.

The invention discloses a multi-task processing system for a large language model, and the system comprises an image input module which is used for receiving to-be-processed image data, can support a plurality of common image formats, including but not limited to JPEG and PNG formats, and converts an image into an array format suitable for subsequent processing through an image reading library; and the image preprocessing subsystem is composed of a normalization unit and a feature extraction unit, and the normalization unit is used for carrying out normalization processing on pixel values of the input image. According to the system, through a series of logically coherent steps of image preprocessing, task coding, multi-task processing and result integration and post-processing, the problem of multi-task processing is solved more systematically. In the aspect of efficiency, through unified preprocessing and task coding, the capacity of a large language model can be better utilized, unnecessary calculation is reduced, therefore, consumption of hardware resources and processing time can be possibly reduced, and the accuracy and usability of results are improved through the steps of result integration and post-processing.

The invention discloses a watermark embedding and extracting method and system based on a Swindow-Unet neural network, and the method comprises the steps: S1, carrying out the shallow local feature extraction of a carrier image and a watermark image through a multi-scale convolution module and an SGE grouping space attention mechanism, and generating a shallow feature map; s2, based on the shallow layer feature map, deep global feature extraction is carried out through a Swinin-Unet encoder of a U-shaped symmetrical structure, the extracted carrier image deep features and watermark image deep features are spliced in the channel direction, feature fusion is carried out through a linear layer, and a watermark-containing image is generated; s3, performing noise attack processing on the watermark-containing image, simulating a real attack scene, and generating a noise image; s4, discriminating the noise image through a double-discriminator structure, and adjusting the starting state of a Gaussian high-pass filter in a watermark embedding discriminator according to a noise type judgment result output by the JPEG noise discriminator; and S5, based on the noise image, extracting a watermark image through a Swinin-Unet decoder which is symmetrical to the encoder in structure.

The application discloses a satellite image compression method under extremely low bandwidth conditions. The steps are as follows: collecting a static image dataset; establishing a variable code rate image coding and decoding network; training the coding and decoding network; embedding a generative adversarial network at the decoding end and performing retraining; embedding a super-resolution network after the coder and decoder and performing retraining; a user selects a region of interest and compresses the selected region; the decoding end restores the code stream into an image and displays the image; the user's unselected region is compressed; when bandwidth is sufficient, the unselected region is transmitted, and the decoding end splices the image; and super-resolution processing is performed. The image transmission method can provide better subjective effect reconstruction effect at the same code rate in the same comparison of a large number of test pictures under the limitation of extremely low bandwidth, compared with traditional image coding such as JPEG and deep learning-based image coding methods.

The application provides a JPEG image partial modification method and device, which comprises the following steps: receiving data from a JPEG data stream, processing the data bit by bit, and reading the data into a 256-byte input buffer in sequence; receiving a JPEG header, decoding minimum coding units (MCU) of the image in sequence according to parameters read from the header, and determining whether the current MCU needs to be modified by comparing whether the coordinates of the current MCU are in a specified image region to be modified; if the current MCU needs to be modified, decoding the current MCU into YUV data, obtaining a character dot matrix corresponding to the current MCU from a character library, superimposing the character dot matrix onto the YUV data, re-encoding the YUV data superimposed with the character dot matrix, and writing the YUV data into an output buffer; and if the current MCU does not need to be modified, directly writing the current MCU into the output buffer. Through the application, image data can be quickly and stably modified and transmitted, and data loss caused by image data delay and loss is reduced.

The present application relates to the technical field of image processing, and particularly relates to a heterogeneous fidelity compression method based on image data clustering. The present application provides a heterogeneous fidelity compression method based on image data clustering, comprising: performing high-low bit depth splitting on an input infrared image to obtain high-bit data and low-bit data; performing JPEG-LS lossless compression on the high-bit data and H.264 intra-frame lossy compression on the low-bit data to obtain a first compressed code stream and a second compressed code stream respectively; and uniformly packaging the first compressed code stream and the second compressed code stream to form a unified compressed stream, wherein the unified compressed stream is transmitted or stored. The present scheme can simultaneously meet the requirements of high compression ratio, key gray scale fidelity and embedded real-time implementation in infrared image compression.

The application discloses a color image quantization step estimation method based on frequency clustering prior knowledge, first, the color image obtained in advance is pretreated, and the image is converted from a spatial domain to a frequency domain; second, an improved Res2Net-C network structure is constructed to obtain quantization step information in the frequency domain; finally, a multi-channel convolution is introduced to assist in estimating the quantization step of the chroma channel. The application converts the image from the spatial domain to the frequency domain through the pretreatment operation, facilitates the network to explore the traces of the quantization step in the JPEG image, and significantly improves the accuracy of estimating the quantization step; the new Res2Net-C network structure can explore the multi-scale information of the image, thereby improving the accuracy of estimating the quantization step; compared with the traditional method for estimating the quantization step, the application has high accuracy and is easier to train.

The application discloses a video stream compression and decompression method and system based on OpenMP thread nesting, adopts parallel pipeline to process video frame data, simplifies original JPEG coding, compresses RGB data into YCbCr data for transmission, adopts OpenMP thread nesting to process the compression and decompression of the video frame data in a multithreading mode under the condition that the outer thread processes the video frame data by using the parallel pipeline, effectively improves the support resolution and fluency of remote playing video, saves the engineering cost of client computing, and improves user experience.

The present application relates to the technical field of document image tamper detection, and particularly relates to a document image tamper detection method and system. The method performs geometric correction on a document image and establishes an orthogonal pixel coordinate system, and extracts key text objects and background regions thereof; based on background texture features, a unified JPEG grid reference point and a high-frequency zero-value mask of the whole document are obtained by statistics, and the construction of a self-reference reference is realized; the grid reference and the zero-value mask are used to extract the frequency domain residual of the text region, and the frequency domain residual is converted into a binary topological graph; the noise dispersion is obtained by analyzing the connected domain distribution, and the feature dimension is converted from the energy amplitude to the geometric morphology; the key objects are grouped based on the layout coordinates, the relative reference of the group is calculated, and the tamper state is judged by comparing the difference between the noise dispersion and the reference. The present application effectively solves the problem of confusion between original content and tamper traces in metadata-free and low signal-to-noise ratio documents, and significantly improves the accuracy and robustness of detection.