A Video Stabilization Technique Based On Optical Flow Sensor

Image vision is an important means of context awareness of the surrounding environment by humans and smart devices,and has been widely used in social life.Among them,image video not only directly serves as the information medium for human beings to understand the world,but also is used as the main way to obtain the state of the surrounding environment by many intelligent devices such as drones,smart ships and self-driving cars.The intelligent device obtains relevant information through pre-processing,analysis and recognition of the video image sequence,and then guides the subsequent motion path planning and state change decision.However,in practical applications,due to complex environmental conditions such as atmospheric turbulence,waves or bumps on the ground,the carrier platform is often shaken and rotated,resulting in instability of the video image sequences.In addition,it seriously affects the video quality,which is not conducive to subsequent image recognition and decision planning.Therefore,in order to improve the video quality,eliminate jitter and rotational motion,it is necessary to study and use a video image stabilization technique.The most direct method in the early days was to use the mechanical image stabilization camera to compensate the motion of the camera equipment,and then proposed a method based on optical image stabilization to reduce the volume of the image stabilization device.However,this method tends to be complicated in structure and has a poor image stabilization effect.With the continuous development of computer and image processing technology,electronic image stabilization technology has gradually become the main research direction of video image stabilization due to its simple structure and low cost.What comes is the decline in application range and image stabilization rate.Aiming at the problems existing in current image stabilization technology,this paper takes electronic image stabilization technology as the research focus,and proposes a video stabilization technology based on optical flow sensor.The image stabilization technology firstly uses a three-axis brushless head to mechanically stabilize the image capturing device;then,by improving the optical flow sensor,it outputs accurate motion vector information about the carrier platform,and finally uses the motion vector to compensate the video image sequence by frame motion,thereby outputing a stable video image sequence.The image stabilization technology applies the high-speed and high-precision characteristics of the optical flow sensor,and avoids the problem that the motion image estimation is complicated and the application environment is limited in the conventional image stabilization technology.The experimental results show that the peak signal-to-noise ratio is improved by 64.99% for pure cloud-level mechanical image stabilization,while the image stabilization time is reduced by 53.93% compared with the traditional electronic image stabilization.Therefore,the video stabilization technology based on the optical flow sensor realizes a beneficial combination with the mechanical image stabilization,and improves the real-time performance,and has practical significance in practical applications.