| Mobile shooting panoramas with large screen,a strong visual impact,more of record and Unparalleled creativity.Aiming at the convenient mobile phone image stitching technology needs,research based on feature phones panoramic image stitching technology.Mobile phone shooting because it is difficult to complete the light heart coincidence,so the use of single-should matrix splicing image can easily lead to dislocation,ghost shadow,serious when it will be difficult to splice images resulting in the termination of the program;SIFT,SURF and other derivative algorithms because of its characteristics of slow speed,patent charges,such as the ORB,such as fast algorithm because of its characteristics too little to meet the needs.This paper gives a solution to the above problems.The key factors such as the quantity and speed of feature extraction in mobile phone shooting,Compared with sift,SURF,Orb,AKAZE,FREAK and other feature extraction algorithms,and Combining the limitations of SIFT,SURF and Orb.A panoramic image mosaic algorithm based on AKAZE(Accelerated-KAZE,Accelerated KAZE)feature detection and FREAK feature description is presented.it can guarantee the number of feature points and increase the computational speed,The experimental results show that the image of horizontal parallax can get good stitching effect.As a result of mobile phone photography panoramic photographs need to keep the mobile phone horizontal movement,and need to move slowly to keep the heart of the coincidence,otherwise there will be ghost,dislocation and other phenomena.Mosaic image overlapping area depth information mismatch is the root of the problem.This paper proposes a combination of improved AKAZE algorithm and RANSAC algorithm to find more matching pairs,combining with APAP algorithm to solve the problem of depth information mismatch.The panoramic image mosaic of non-horizontal parallax is realized.In the end,the image stitching is achieved in the PC platform,and the stitching effect is good. |
PDF Full Text Request