Research Of Video-based Human-Computer Interaction Mode

Nature Human-Computer Interaction is the development direction of future computer technology, the ultimate goal is that computer can see, hear, speak, and think. In the paper, HCI system interacts naturally with the color indicators as the interactive media and camera as the input device, research focuses on the color indicators'detection and tracking, which could be divided into three parts:Firstly, extract object's motion region. If the color indicators are detected in the whole image, it will be badly interfered by the background with similar color. Taking the color indicators in the motion region into consideration, first of all, background image is chosen based on the similarity between consecutive frames of a video in L channel of Lab color space. Then image differentiation on any frame and the background can identify the motion region, which can reduce the search area for the color indicators. The algorithm is simple with low computation and good robustness.Secondly, detect color indicators. The color indicators are affected easily by light condition and background condition, and color threshold shifts, but in the same scenarios, it is stabler. Firstly, find the convex hull of motion region and take samples within it. By assigning the weight, seeds can be determined using clustering method. Secondly, region growth is implemented by applying Bayesian decision with minimal error ratio. The proposed method is applicable to more conditions and contributes to better results compared with traditional color-threshold vector method.Thirdly, track color indicators. Color indicators'tracking implements with Mean Shift algorithm in the following frame according to the results of detection. After finishing the experiment we have found that color indicators lost when it moves fast or system handles fewer frames per second, so there is no overlap in the target area between the adjacent frames. This problem can be overcomed by applying Mean Shift Algorithm based on kalman filter prediction or Mean Shift Algorithm based on Grey prediction. The former has large amount of computation, which only handle fewer frames per second, bringing about not guaranteeing the continuity of target motion in the video. The latter can overcome those limitations, which becomes more suitable.The result shows the color indicators'detection method and tracking algorithm can meet the demand of HCI system, has certain practical value and application prospect.