CN115952139A - Multi-frame three-dimensional image processing method and system for mobile equipment - Google Patents
Multi-frame three-dimensional image processing method and system for mobile equipment Download PDFInfo
- Publication number
- CN115952139A CN115952139A CN202310243477.1A CN202310243477A CN115952139A CN 115952139 A CN115952139 A CN 115952139A CN 202310243477 A CN202310243477 A CN 202310243477A CN 115952139 A CN115952139 A CN 115952139A
- Authority
- CN
- China
- Prior art keywords
- data
- frame
- memory
- dimensional
- pixel
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Abstract
The invention relates to the field of computers, in particular to a multi-frame three-dimensional image processing method and a system for mobile equipment.A three-dimensional multi-frame file is read and analyzed, and is stored into a memory according to a pixel coordinate mode to obtain first data; then calling an arm register to load first data, using a Neon mathematical instruction set to operate the first data, and then storing the first data into a memory to obtain second data; and finally calling an arm register to load second data, and performing rasterization superposition on the second data by using a Neon mathematical instruction set based on a ray projection method to obtain final data and storing the final data into a memory. By adopting the analysis method, the performance of accessing the memory data is greatly enhanced.
Description
Technical Field
The invention relates to the field of computers, in particular to a multi-frame three-dimensional image processing method and a multi-frame three-dimensional image processing system for mobile equipment (a tablet personal computer) of an Arm platform, wherein the multi-frame three-dimensional image processing method and the system are used for processing super-large multi-frame three-dimensional images stored on a cloud server by the mobile equipment (the tablet personal computer) of the Arm platform and performing rendering display after three-dimensional reconstruction.
Background
The 3D reconstruction and rendering of multi-frame three-dimensional images have been implemented and displayed mainly by high-performance PCs in the past, and are increasingly processed on mobile devices (mobile phone tablets) along with the popularization of the mobile devices. The hardware resources and processing power of the mobile device are weak compared to the high performance pc.
In a conventional scheme, a plurality of frames of three-dimensional images are loaded from a cloud server or a local disk, a plurality of frames of data are read and buffered frame by frame into a memory to form a memory array with a frame as a unit, a ray projection method is used for traversing and operating a coordinate pixel area corresponding to each frame in each memory array, and finally two-dimensional rasterized data are formed and displayed (refer to fig. 1).
For example: the prior art discloses an orthographic video image generation method (publication number: CN 114627237A) based on a real-scene three-dimensional model, which explains a TIFF three-dimensional reconstruction method, but does not have an image to an arm platform, and the method does not optimize aiming at an arm end, so that the operation efficiency on the arm platform is not high. The prior art also discloses methods for creating, storing and providing access to three-dimensional scanned images (publication No. 103038780A) where data from the three-dimensional reconstruction stage is stored and indexed through a database, which only considers application logic storage and access simplicity and is not optimized for low latency high performance data processing of arm memory physical layout and TIFF three-dimensional reconstruction.
Disclosure of Invention
The technical problem of the invention is mainly solved by the following technical scheme:
a multi-frame three-dimensional image processing method for mobile equipment is based on an arm platform and comprises the following steps
Reading and analyzing a three-dimensional multi-frame file, and storing the three-dimensional multi-frame file into an internal memory according to a pixel coordinate mode to obtain first data;
calling an arm register to load first data, using a Neon mathematical instruction set to operate the first data, and then storing the first data into a memory to obtain second data;
and calling an arm register to load second data, and performing rasterization superposition on the second data by using a Neon mathematical instruction set based on a ray projection method to obtain final data to be stored in a memory.
In the above-mentioned multi-frame three-dimensional image processing method for a mobile device,
reading a file header of a three-dimensional multi-frame file and applying for preprocessing a memory;
and reading pixel data of the three-dimensional multi-frame file in parallel and storing the pixel data into a preprocessing memory.
In the above-mentioned multi-frame three-dimensional image processing method for a mobile device,
calculating the size of a memory block to be allocated according to multi-frame three-dimensional file information, the size of an ARM system memory and the number of light projected by light;
and dividing the applied memory block into memory subblocks according to the number of rays to obtain the memory block consisting of a plurality of memory subblocks.
In the above multi-frame three-dimensional image processing method for a mobile device,
defining the file frame number of a multi-frame file as nFrame, the number of bytes occupied by each pixel point as nBitCount, the number of rays projected by light as nRay, the number of pixel points sampled by each light as nPixel, and the memory size of an ARM system as nDDR, and then applying for the memory size nMEM = nFrame nBitCount nDDR nPixel with 4 bytes alignment in advance.
In the above method for processing multi-frame three-dimensional images of a mobile device, the number nmeccont = nMem/nRay of a plurality of memory subblocks, and the head address of each pre-allocated memory subblock is stored in the cache.
In the above-mentioned multi-frame three-dimensional image processing method for a mobile device,
reading pixel data of a multi-frame file in parallel according to an openmp module of an ARM platform, wherein the pixel data processing specifically comprises first-stage pixel data processing and second-stage pixel data processing, and the first data is pixel data after the second-stage processing;
and saving the first data to a pre-allocated memory block.
In the above-mentioned multi-frame three-dimensional image processing method for a mobile device,
the first-stage pixel data processing is that each frame of data is circularly traversed;
and the second-stage pixel data processing is to store sampling rectangular pixels of the light projection method into a pre-allocated memory according to the direction of increasing horizontal and vertical coordinates in each frame of data in a circulating manner.
In the foregoing method for processing a multi-frame three-dimensional image of a mobile device, the number of rays projected by a ray is defined as nRay =4, the resolution of a single frame is 2 × 2, the number of pixels sampled by each ray is nPixel =1, the number of frames of a multi-frame three-dimensional file is nFrame =2, the number of bytes occupied by each pixel is nBitCount, the starting address of a memory block is nddrraddrinit, and the storage address of the nth pixel of the mth frame in the memory is nddrraddr = nddrinit + (nPixel nRay × n + m) × nBitCount = ndraddrinit + (4n + m) × ntcount, where 0 + m is nFrame, and 0 + n is truss 4.
In the above-mentioned multi-frame three-dimensional image processing method for a mobile device,
loading the first data into an ARM Neon double-word vector register group;
performing mathematical operation in a ray projection algorithm by using a Neon intunistics instruction set, and storing the processed data of each frame after rectangular processing into a memory to obtain second data; and loading the second data into an ARM Neon double-word vector register group, rasterizing the register group, uniformly superposing the data of the total nFrame frame number once by using a Neon instruction, and writing the superposed data into a memory to obtain the final three-dimensional reconstructed image data.
A mobile device multi-frame three-dimensional image processing system is characterized by comprising:
a first module: the device comprises a memory, a three-dimensional multi-frame file reading module, a three-dimensional multi-frame file analyzing module, a pixel coordinate module and a data processing module, wherein the three-dimensional multi-frame file reading module is configured for reading and analyzing three-dimensional multi-frame files in parallel and storing the three-dimensional multi-frame files into the memory according to a pixel coordinate mode to obtain first data;
a second module: the data processing method comprises the steps that an arm register is called to load first data, a Neon mathematical instruction set is used for calculating the first data and then storing the first data into a memory, and second data are obtained;
a third module: and the system is configured to call an arm register to load second data, and perform rasterization superposition on the second data by using a Neon mathematical instruction set based on a ray projection method to obtain final data to be stored in a memory.
Therefore, the invention has the following advantages: 1. in the three-dimensional multi-frame file reading stage, concurrent processing is performed by using an arm openmp technology, so that the reading and resolving performance of the cloud server file is accelerated. 2. The read multi-frame data is stored in a memory according to the pixel coordinates, and is not stored in a frame unit, so that the processing performance of the subsequent ray projection method is accelerated, and the performance of accessing the memory data is greatly enhanced. 3. The production line of the light projection method for rasterization operation uses the armneon technology for parallel acceleration. 4. The nero instruction is used for increasing the processing bandwidth and reducing the occupancy rate of the multi-core cpu. 5. After the memory access layout is optimized, the total amount of physical memory used is reduced. And 6, after the CPU and the memory resource are used, the ecological compatibility of the fragmented mobile equipment is better.
Drawings
FIG. 1 is a schematic flow diagram of a prior art process.
FIG. 2 is a schematic flow diagram of the process of the present invention.
Fig. 3 is a diagram illustrating a layout of a multi-frame data memory according to the prior art.
Fig. 4 is a schematic diagram of a multi-frame dataram layout according to the present invention.
FIG. 5 is a schematic diagram of a TIFF file loading process.
FIG. 6 is a schematic diagram of a conventional ray-casting sequential process.
FIG. 7 is a schematic diagram of an algorithm acceleration process of ray casting using ARM Neon technology.
Detailed Description
The technical scheme of the invention is further specifically described by the following embodiments and the accompanying drawings.
Example (b):
the following description specifically explains an example in which a plurality of three-dimensional images are TIFF files.
1. Firstly, the whole process of analyzing the TIFF file in the prior art is explained, and the process comprises the following steps:
1. reading a TIFF file header, analyzing a DE (Directory Entry) Directory Entry, and obtaining the offset of each frame in the multi-frame data relative to the file header according to the Directory Entry;
2. reading single-frame file data from a first frame in sequence, analyzing pixels (RGB) of each coordinate, and storing the analyzed pixels into a memory, wherein the memory is directly applied from a pile before use, and the final memory layout is sequential X X Y nBit n, wherein X = the image width, Y = the image height, nBit = the number of bytes (generally 3 or 4) stored in each pixel RGB, and n = the number of frames;
3. and carrying out ray projection operation and combination on the same pixel region of the multi-frame images.
2. The invention improves the data processing links, and comprises the following detailed steps:
1. and reading a file header of the TIFF image file, and applying for preprocessing the memory for a multi-frame image light projection method. Because the ray casting operation process needs a large amount of access to the memory, the memory buffer design needs to improve the access performance as much as possible. The execution steps are as follows: A. according to the TIFF file information, the ARM system memory size, the ray number of ray projection and the size of a memory block to be allocated, assuming that the frame number of the TIFF file is nFrame, the number of bytes occupied by each pixel point is nBitCount, the ray number of the ray projection is nRay, the pixel point number of each ray sampling is nPixel, and the ARM system memory size is nDDR, applying for the memory size nMEM = nTracame nBitCount nDDR nPixel with 4 byte alignment in advance. Generally, the nMem should be up to 50% of the nDDR, i.e., ARM system memory, and excessive applications will degrade other system service performance.
B. Dividing the applied memory block into subblocks according to the number of rays, wherein the number of subblocks nMemCount = nMem/nRay, the first address of each subblock needs to be 64 bytes aligned, and using an ARM PLD command to be prestored in the cache (the 64 bytes are aligned to be correct for the execution of the PLD command, and the access requirement of the cache line is 64 bytes aligned address access). Therefore, when the memory data operation is subsequently carried out, the initial address of the memory block can be directly obtained from the cache, and the read-write operation is faster.
Analyzing the TIFF multi-frame file, comprising the following steps of:
A. the pixel data of the multi-frame file is read in parallel by using an openmp technology of an ARM platform, and the TIFF super-large file is stored in a cloud end, so that parallel access is supported in a mechanism. Because the cloud data is stored in a distributed manner, parallel access to different data segments of the same file can be distributed to different processing network storage areas to perform reading and writing. The parallel processing design is as follows (see fig. 5):
a. opening a cloud file;
b. the first-stage circulation traverses each frame of data, and the second-stage circulation stores sampling rectangular pixels of a ray projection method into a pre-allocation memory in each frame of data according to the direction of increasing horizontal and vertical coordinates;
c. since there is no correlation between frames and the pixel data in a single frame is correlated, the parallel processing of nested 2-level openmp is used, the single frame data combing is processed at the first level, and the pixel data processing in the frame is processed at the second level.
B. And (4) storing the pixel data after each frame of data is analyzed into the memory pre-allocated in the step (1). According to 2 frames of images, 4 pixels per frame of image, the ray projection ray processing granularity is a simple model of one pixel, the image data in the prior art is used for organizing the memory data according to a frame sequence of text TIFF (time series of fusion transform) elements (refer to FIG. 3), while the memory data in the invention is organized according to the ray penetration sequence of the ray projection method (refer to FIG. 4), the memory access efficiency of ray projection by the memory layout is higher, because the algorithm processing is operated and overlapped according to the same pixel area of each frame, namely the same area of all frames is processed once, and then the next area is switched to for processing.
3. The arm neon technique is used to optimize the ray-casting process flow.
Even if the data is buffered in the pre-allocated memory, the data can be accessed efficiently, the calculation amount is large, the data access processing is high, and the optimization can be performed by using arm neon through an improved algorithm.
In the prior art, the ray-casting processing flow is to sequentially perform operations and read-write operations on memory data (refer to fig. 6).
The invention uses the arm neon instruction set to reduce parallel memory reading and writing, because the performance is improved by more than 5 times by using a 128-bit arm register compared with using 128-bit memory for access.
The specific implementation steps are as follows (see fig. 7):
A. loading pixel set data of a frame rectangular block in a memory into an ARM Neon double-word vector register group;
B. the Neon intubatics instruction set is used for carrying out mathematical operations in the ray projection algorithm, and the operations are directly completed by using the Neon mathematical instruction set without using a conventional ARMRISC instruction set. After the processing is finished, the data after each frame of rectangular processing needs to be stored in a memory, and the ARM Neon register has limited space, cannot store all data and must be buffered in the memory;
C. the final step of the projection method is to rasterize the rectangular calculation data of all frames, i.e. to superimpose the data of the total number of nFrame frames once. The processing also comprises the steps that data in the memory are loaded into a Neon register set at present, the Neon instruction is directly used for superposition, and the superposed data are written into the memory.
Overall, although the use of the Neon register increases the copy between the memory and the register for 2 times, this overhead is greatly improved compared with the performance of accessing the memory for many times and performing mathematical operation by using the conventional ARM risc instruction set, the former only has the first and last 2 times of registers and memory to transmit data, other instruction operations are directly performed at the register level without being completed, and the latter performs the operations in sequence each time: the internal access instruction- > the internal memory data is loaded into the register- > the register is calculated by using the risc instruction set- > the register data is written into the internal memory.
In this embodiment, a multi-frame three-dimensional image processing system for a mobile device is further provided, including:
a first module: the device comprises a memory, a three-dimensional multi-frame file reading module, a three-dimensional multi-frame file analyzing module, a pixel coordinate module and a data processing module, wherein the three-dimensional multi-frame file reading module is configured for reading and analyzing three-dimensional multi-frame files in parallel and storing the three-dimensional multi-frame files into the memory according to a pixel coordinate mode to obtain first data;
a second module: the device comprises an arm register, a memory and a data processing unit, wherein the arm register is configured to call the arm register to load first data, and the first data is operated by using a Neon mathematical instruction set and then is stored in the memory to obtain second data;
a third module: and the second data are loaded by calling an arm register, and the final data are obtained by performing rasterization superposition on the second data by using a Neon mathematical instruction set based on a ray projection method and then stored in a memory.
3. The analytic method of the invention is applied to the following concrete configuration: in the ARM Cortex-a76+ linux 5.10 version and 8G memory devices, the parameters for storing TIFF files on the cloud server are as follows: epfl _ training. Tif/file size: 123.90 MB/resolution: 1024 726 165/165 frames; opening a file on an ARM system by using the prior art scheme and rendering the file by using a VTK light projection method, wherein the total duration is 12 seconds; by using the optimization scheme of the invention, the final total market of the finished rendering is 5 seconds, and the promotion is nearly 120%.
The specific embodiments described herein are merely illustrative of the spirit of the invention. Various modifications or additions may be made to the described embodiments or alternatives may be employed by those skilled in the art without departing from the spirit or ambit of the invention as defined in the appended claims.
Claims (10)
1. A multi-frame three-dimensional image processing method for mobile equipment is characterized by comprising the following steps based on an arm platform
Reading and analyzing a three-dimensional multi-frame file, and storing the three-dimensional multi-frame file into an internal memory according to a pixel coordinate mode to obtain first data;
calling an arm register to load first data, using a Neon mathematical instruction set to operate the first data, and then storing the first data into a memory to obtain second data;
and calling an arm register to load second data, and performing rasterization superposition on the second data by using a Neon mathematical instruction set based on a ray projection method to obtain final data and storing the final data into a memory.
2. The multi-frame three-dimensional image processing method of the mobile device according to claim 1,
reading a file header of a three-dimensional multi-frame file and applying for preprocessing a memory;
and reading the pixel data of the three-dimensional multi-frame file in parallel and storing the pixel data into a preprocessing memory.
3. The multi-frame three-dimensional image processing method of the mobile device according to claim 2,
calculating the size of a memory block to be allocated according to multi-frame three-dimensional file information, the size of an ARM system memory and the number of light projected by light;
and dividing the applied memory block into memory subblocks according to the number of rays to obtain the memory block consisting of a plurality of memory subblocks.
4. The multi-frame three-dimensional image processing method of the mobile device according to claim 3,
defining the file frame number of a multi-frame file as nFrame, the number of bytes occupied by each pixel point as nBitCount, the number of rays projected by the rays as nRay, the number of pixel points sampled by each ray as nPixel, and the size of an ARM system memory as nDDR, and then applying for the memory size nMEM = nFrame nBitCount nDDR nPixel with 4 byte alignment in advance.
5. The multi-frame three-dimensional image processing method of the mobile device according to claim 4, wherein the number of memory subblocks nMemCount = nMem/nRay, and a pre-allocated head address of each memory subblock is stored into the cache.
6. The multi-frame three-dimensional image processing method of the mobile device according to claim 1,
reading pixel data of a multi-frame file in parallel according to an openmp module of an ARM platform, wherein the pixel data processing specifically comprises first-stage pixel data processing and second-stage pixel data processing, and the first data is pixel data after the second-stage processing;
and saving the first data to a pre-allocated memory block.
7. The multi-frame three-dimensional image processing method of the mobile device according to claim 6,
the first-stage pixel data processing is that each frame of data is circularly traversed;
and the second-stage pixel data processing is to store sampling rectangular pixels of the light projection method into a pre-allocated memory according to the direction of increasing horizontal and vertical coordinates in each frame of data in a circulating manner.
8. The method as claimed in claim 1, wherein the number of rays for ray projection is nRay =4, the resolution of a single frame is 2 × 2, the number of pixels for each ray sampling is nPixel =1, the number of frames of the multi-frame three-dimensional file is nFrame =2, the number of bytes occupied by each pixel is nBitCount, the starting address of the memory block is nddrraddrinit, and the storage address of the nth pixel of the mth frame in the memory is nddrraddrinit = nddrdrdrdrinit + (nPixel nRay × n + m) = nBitCount = ndradradinit + (4n + m) ntbcount, where 0< m < + >, nFrame,0< n 4.
9. The multi-frame three-dimensional image processing method of the mobile device according to claim 1,
loading the first data into an ARM Neon double-word vector register group;
performing mathematical operation in a ray projection algorithm by using a Neon intunistics instruction set, and storing the processed data of each frame after rectangular processing into a memory to obtain second data; and loading the second data into an ARM Neon double-word vector register group, rasterizing the register group, uniformly superposing the data of the total nFrame frame number once by using a Neon instruction, and writing the superposed data into a memory to obtain the final three-dimensional reconstructed image data.
10. A multi-frame three-dimensional image processing system of a mobile device is characterized by comprising:
a first module: the device comprises a memory, a storage, a data processing module and a data processing module, wherein the memory is configured to be used for reading and analyzing three-dimensional multi-frame files in parallel and storing the three-dimensional multi-frame files into the memory according to a pixel coordinate mode to obtain first data;
a second module: the data processing method comprises the steps that an arm register is called to load first data, a Neon mathematical instruction set is used for calculating the first data and then storing the first data into a memory, and second data are obtained;
a third module: and the system is configured to call an arm register to load second data, and perform rasterization superposition on the second data by using a Neon mathematical instruction set based on a ray projection method to obtain final data to be stored in a memory.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310243477.1A CN115952139B (en) | 2023-03-14 | 2023-03-14 | Multi-frame three-dimensional image processing method and system for mobile equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310243477.1A CN115952139B (en) | 2023-03-14 | 2023-03-14 | Multi-frame three-dimensional image processing method and system for mobile equipment |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115952139A true CN115952139A (en) | 2023-04-11 |
CN115952139B CN115952139B (en) | 2023-06-30 |
Family
ID=85891390
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310243477.1A Active CN115952139B (en) | 2023-03-14 | 2023-03-14 | Multi-frame three-dimensional image processing method and system for mobile equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115952139B (en) |
Citations (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100103282A1 (en) * | 2008-10-28 | 2010-04-29 | Kabushiki Kaisha Toshiba | Image processing apparatus and image processing system |
US20140081918A1 (en) * | 2012-09-18 | 2014-03-20 | Mapr Technologies, Inc. | Table format for map reduce system |
CN103971396A (en) * | 2014-05-24 | 2014-08-06 | 哈尔滨工业大学 | OpenGL ES (open graphics library for embedded system) implementation method for ray casting algorithm under ARM+GPU (advanced RISC machine+graphic processing unit) heterogeneous architecture |
CN108564652A (en) * | 2018-03-12 | 2018-09-21 | 中国科学院自动化研究所 | Efficiently utilize the high-precision three-dimensional method for reconstructing of memory and system and equipment |
US20190026243A1 (en) * | 2017-07-18 | 2019-01-24 | Arm Limited | Display controller |
CN111950404A (en) * | 2020-07-29 | 2020-11-17 | 南京大学 | Single-image three-dimensional reconstruction method based on deep learning video surveillance |
US20200372710A1 (en) * | 2019-05-24 | 2020-11-26 | Adobe, Inc. | 3d object reconstruction using photometric mesh representation |
US20210067766A1 (en) * | 2019-08-30 | 2021-03-04 | Beijing Boe Optoelectronics Technology Co., Ltd. | Data transmission method, method of displaying three-dimensional image, data transmission device and three-dimensional image display device |
CN113424232A (en) * | 2019-12-27 | 2021-09-21 | 深圳市大疆创新科技有限公司 | Three-dimensional point cloud map construction method, system and equipment |
CN113597603A (en) * | 2020-10-14 | 2021-11-02 | 深圳市大疆创新科技有限公司 | Data processing method, device, equipment and storage medium |
CN113628304A (en) * | 2021-10-09 | 2021-11-09 | 湖北芯擎科技有限公司 | Image processing method, image processing device, electronic equipment and storage medium |
WO2022111619A1 (en) * | 2020-11-30 | 2022-06-02 | 华为技术有限公司 | Image processing method and related apparatus |
-
2023
- 2023-03-14 CN CN202310243477.1A patent/CN115952139B/en active Active
Patent Citations (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100103282A1 (en) * | 2008-10-28 | 2010-04-29 | Kabushiki Kaisha Toshiba | Image processing apparatus and image processing system |
US20140081918A1 (en) * | 2012-09-18 | 2014-03-20 | Mapr Technologies, Inc. | Table format for map reduce system |
CN103971396A (en) * | 2014-05-24 | 2014-08-06 | 哈尔滨工业大学 | OpenGL ES (open graphics library for embedded system) implementation method for ray casting algorithm under ARM+GPU (advanced RISC machine+graphic processing unit) heterogeneous architecture |
US20190026243A1 (en) * | 2017-07-18 | 2019-01-24 | Arm Limited | Display controller |
CN108564652A (en) * | 2018-03-12 | 2018-09-21 | 中国科学院自动化研究所 | Efficiently utilize the high-precision three-dimensional method for reconstructing of memory and system and equipment |
US20200372710A1 (en) * | 2019-05-24 | 2020-11-26 | Adobe, Inc. | 3d object reconstruction using photometric mesh representation |
US20210067766A1 (en) * | 2019-08-30 | 2021-03-04 | Beijing Boe Optoelectronics Technology Co., Ltd. | Data transmission method, method of displaying three-dimensional image, data transmission device and three-dimensional image display device |
CN113424232A (en) * | 2019-12-27 | 2021-09-21 | 深圳市大疆创新科技有限公司 | Three-dimensional point cloud map construction method, system and equipment |
CN111950404A (en) * | 2020-07-29 | 2020-11-17 | 南京大学 | Single-image three-dimensional reconstruction method based on deep learning video surveillance |
CN113597603A (en) * | 2020-10-14 | 2021-11-02 | 深圳市大疆创新科技有限公司 | Data processing method, device, equipment and storage medium |
WO2022111619A1 (en) * | 2020-11-30 | 2022-06-02 | 华为技术有限公司 | Image processing method and related apparatus |
CN113628304A (en) * | 2021-10-09 | 2021-11-09 | 湖北芯擎科技有限公司 | Image processing method, image processing device, electronic equipment and storage medium |
Non-Patent Citations (1)
Title |
---|
张燕: "骨科CT数据的图像预处理及三维重建技术的研究", 中国优秀硕士学位论文全文数据库信息科技辑 * |
Also Published As
Publication number | Publication date |
---|---|
CN115952139B (en) | 2023-06-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
EP3667607B1 (en) | Virtual linebuffers for image signal processors | |
US6825847B1 (en) | System and method for real-time compression of pixel colors | |
JP6352546B2 (en) | Handling unaligned block transfer operations | |
JP2004518343A (en) | Apparatus and method for memory address translation and image processing apparatus including such an apparatus | |
US9304933B2 (en) | Techniques to request stored data from a memory | |
US20060176310A1 (en) | Method and apparatus for de-indexing geometry | |
US9934547B2 (en) | Method and system for reducing the number of draw commands issued to a graphics processing unit (GPU) | |
US11348203B2 (en) | Image generation using subscaling and depth up-scaling | |
WO2021248705A1 (en) | Image rendering method and apparatus, computer program and readable medium | |
US20160189417A1 (en) | Method, apparatus and system for rendering an image | |
US10070134B2 (en) | Analytics assisted encoding | |
CA2558657A1 (en) | Embedded system with 3d graphics core and local pixel buffer | |
US9607417B2 (en) | Parallel rendering of region-based graphics representations | |
CN105247478A (en) | Command instruction management | |
JP2012242855A (en) | Data processing apparatus and data processing method | |
EP2453360B1 (en) | Method and apparatus for translating memory access address | |
CN115952139B (en) | Multi-frame three-dimensional image processing method and system for mobile equipment | |
US20230297388A1 (en) | Method and system to process asynchronous and distributed training tasks | |
US7490208B1 (en) | Architecture for compact multi-ported register file | |
CN105427235A (en) | Image browsing method and system | |
US6496185B1 (en) | Method and apparatus for processing a mesh of triangles | |
US8810572B2 (en) | Tessellation cache for object rendering | |
US8941881B2 (en) | Method and apparatus for rasterizing transparent page | |
CN115564644B (en) | Image data processing method, related device and computer storage medium | |
CN112233206B (en) | Character image generation method, device, equipment and storage medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |