Realization Of Image Stitching Technology Based On FPGA

Image splicing is an important branch of digital image processing,it means that making a number of images with overlapping areas seamlessly spliced into a larger image by using image registration and image fusion.The image splicing can make multiple images which captured in the same scene but with different viewpoints displayed as one single image with high definition and large viewpoint.As a result,the conflicts between the multiple-collection and single-display,narrow viewpoint-collection and wide viewpoint-display are avoided.Therefore,the image splicing technology is widely used in the large-screen display,video conferencing systems,traffic network monitoring,security monitoring and medical imaging.In this paper,the practical significance,research status and development trend of the image splicing technology are firstly introduces,then,the basic flow and key techniques in the image splicing is analyzed.After that,the overall design scheme of the FPGA-based picture splicing technology is systematically explained,and the feasibility is analyzed.The design first stores the two images of 240×180 in the LPM?ROM of FPGA internal chip in the same environment different shooting angle captured;Then extractes 3×3 window as the mathematical model,and conducts the median filtering algorithm,improved Sobel edge detection,image registration based on Harris corner feature,and weighted average image fusion processing;Finally,the two images are combined into a composite image with a large angle of view.Among them,3×3 frame window extraction,median filtering,edge detection and other functional modules are programmed by Verilog HDL language,Harris image registration and image fusion processing algorithm are use the Celoxica DK design tools to achieve the implementation of the algorithm Handle C source compiled into the V program can be implemented in the FPGA and then be called in Quartus II.This design uses Cyclone IV EP4CE17FC8 as the core processing chip,and the top-level file of the whole design is tested and verified on the hardware platform by compiling and simulating.Verification results show that:(1)In the process of image edge detection,by increasing two diagonal direction operator templates in the classical Sobel edge detection algorithm and using the median filter operation,a better effect of edge detection can be achieved.Compared with the classical Sobel algorithm,the advanced Sobel algorithm can better extract the edge information for the image with complicated vein.(2)In the process of image registration,the registration algorithm based on Harris is employed to achieve corner detection for the edge detected image,and then filter out the sub-quality edges and obtain the highly matched pixels.The spatial conversion parameters can be obtained and the original image and the image wait for registration can be jointed based on these information.(3)In the phase of image fusion,the fusion algorithm based on homography matrix is adopted and implemented.The processed image is smooth and natural,and there is no obvious stitching mark in the overlap area.The seamless stitching of the image is realized basically.The FPGA-based image splicing technology in this paper takes full advantage of the high speed parallel data processing ability of the development of FPGA technology and flexible design,and realized the seamless splicing of multiple images.It plays an important role in high-speed image processing and real-time image display and so on.