Research On Image Blending Technology For SOPC Implementation

Image blending plays an important role in the field of high-resolution acquisition and wide-field video processing.However,most image blending methods are based on software implementation,which is difficult to meet the needs of real-time video processing.Taking advantage of the hardware and software co-design of system on programmable chip(SOPC),this thesis proposes a fast image blending method for SOPC implementation,which realizes blending by controlling the storing addresses of source pixels.The starting storing location is accurately computed according to the affine matrix by software,while other storing addresses are computed by hardware circuit on the basis of starting storing location.Based on the Bresenham algorithm in computer graphics and the advantages of burst mode operation of a dynamic memory,analyzing the cause of the seam,this thesis proposes the seam-based variable-step Bresenham(SVSB)method.It simplifies the affine matrix operations to integer addition and subtraction,and thus the computational complexity and storage resource consumption are reduced,which is suitable for hardware circuit implementation.Meanwhile,by using several variable-size steps instead of a constant step to compute the storing addresses,the image quality is greatly improved.The peak signal to noise ratio(PSNR)can increase 5-7dB compared with the constant-step Bresenham method.The blending system is built on the Xilinx field programmable gate array(FPGA)development board,and the type is XUPV5-LX110 T.The software calculates the initial blending parameters and controls the workflow of the whole system.The hardware realizes the synchronization of camera data,color space conversion,arbitration and image blending.This thesis introduces the overall architecture and work flow,analyzes the clock domain and bus bandwidth,and plans the data buffer and registers in detail.Experiments show that the blending system based on SOPC can support real-time blending of dual cameras of 720×576@50fps,with less resource consumption?...