Font Size: a A A

Research On Electronic Image Stabilization

Posted on:2017-01-15Degree:MasterType:Thesis
Country:ChinaCandidate:L J LaiFull Text:PDF
GTID:2278330503478884Subject:Detection Technology and Automation
Abstract/Summary:PDF Full Text Request
With the development of hardware and software, digital image stabilization as a new stably technology is widely used as a basic video enhance technology. With the widely application of the hardware and software and in the advancement of miniaturization of electronic equipment,digital image stabilization has become a new research hotpot. The aim of digital image stabilization is to remove camera jitter to enhance the video quality for observation and computer processing. The essence of digital image stabilization is image motion compensation. The global motion vector is estimated directly from the image information. Obtained global motion vector we will separate expected trajectory and unexpected trajectory. The new stable video is compensated from the unexpected trajectory. The main sacrifice of our article is as follows:As for phase correlation can only deal with translate between two frames, we introduce log polar into phase correlation. The improved method can deal with translation, rotation and scale at the same. The experiment shows that the accuracy of the improved method can reach 0.7 degrees or less.We combine harris features with optical flow and design an image stabilization algorithm based on sparse optical flow. The first step is to obtain harris corners. Here, in order to limit the quantity of harris corners and uniform distribution of harris corner in the image,we add a distance limit mechanism. The we use optical flow to match the harris corners. Optical flow can only works well in tiny movement, in order to improve the performance of optical flow, we add pyramid technology into optical flow. Then using Newton-Raphson in inner-frame and out-frame.Then we obtain the affine parameters using iterative least squares fitting.In motion filtering part, Kalman filtering is used to separate unexpected trajectory. The new stable video compensates from the unexpected trajectory. The experiment shows that the design algorithm has a good effect in removing complex jitter.
Keywords/Search Tags:digital image stabilization, global motion estimation, motion filtering, optical flow, pyramid, Harris
PDF Full Text Request
Related items