• Figure 8-2: System Design Architecture
  • Ultra fast cnn based Hardware Computing Platform Concepts for adas visual Sensors and Evolutionary Mobile Robots




    Download 3,22 Mb.
    Pdf ko'rish
    bet57/81
    Sana16.05.2024
    Hajmi3,22 Mb.
    #238917
    1   ...   53   54   55   56   57   58   59   60   ...   81
    Bog'liq
    Alireza Fasih

    8.5
     
    System Diagram 
    The system diagram gives the clear understanding of the work we have done along with 
    the flow of data and processing steps. The system diagram of the designed model of our 
    research was show in Figure 8-2. 
    Figure 
    8-2: System Design Architecture 
    From the system diagram shown above we can understand that there are several available 
    platforms which can help in the processing of image processing tasks. They are FPGAs, 
    DSPs, CPUs and GPUs. Out of which the OpenCL framework is capable of developing these 
    image processing algorithms on the multi core CPUs and on GPU or on cluster of GPUs. So 
    we have chosen this OpenCL framework as a programming language for our task. This 
    OpenCL framework has an 
    Application Programming Interface
    (API) which helps the 
    processors either CPU or GPU to communicate with the shared memory and image 
    processing unit. This helps in acting as an interface to both of them. The Image processing 
    unit has access to the global memory as a temporary buffer for the processing of images. 
    There is a high level script interpreter for task management and accessing to I/O’s and


     
    91 
    digital channels such as files, Camera, etc. The data which is to be processed is kept in the 
    shared memory and is accessible by the OpenCL framework and the Image processing unit. 
    After loading the instructions from the image processing unit, the API of the OpenCL will 
    try to take the required kernel from the kernel bank, for the process of the instructions.
    After getting the required kernels from the kernel bank the OpenCL make these kernels to 
    be executed on the necessary processor and then the results are again stored in the shared 
    memory which is collected by the image processing unit. The OpenCL API also checks the 
    processing commands and the synchronization commands from the image processing unit 
    in order to perform a correct operation on the proper image. 

    Download 3,22 Mb.
    1   ...   53   54   55   56   57   58   59   60   ...   81




    Download 3,22 Mb.
    Pdf ko'rish

    Bosh sahifa
    Aloqalar

        Bosh sahifa



    Ultra fast cnn based Hardware Computing Platform Concepts for adas visual Sensors and Evolutionary Mobile Robots

    Download 3,22 Mb.
    Pdf ko'rish