Real-Time Stereoscopic Image Generation from Depth Map and Its Hardware Design

• Main Authors: Yu-Wei Chang, 張祐維
• Other Authors: Fang-Hsuan Cheng
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/40614297081644157655

碩士 === 中華大學 === 資訊工程學系(所) === 98 === When we look at the object, the perceived object will cause some displacement in our eyes. This displacement is called parallax. The far the object is, the small the parallax is, and vice versa. Because of the parallax, people have feeling to stereoscopic vision. We can utilize some methods to transform the 2D image to virtual stereoscopic image. For example, DIBR (depth-image-based rendering) algorithm utilizes the depth map and virtual stereoscopic camera model of 2D image to produce the stereoscopic image with parallax. The stereoscopic image is produced by left and right shifting the original image which will cause the hole problem in the image. Therefore an image inpainting algorithm is used to fill the holes. However it is a time consuming operations. In order to achieve the real-time application, a hardware architecture is developed to generate the whole stereoscopic image. In this paper, we utilize the DIBR algorithm to generate the binocular image where a simple image inpainting algorithm is used to fill the hole. In order to increase the computing speed, the whole image is cut into pieces and each piece of image is processed separately. It is found that this kind of processing method can improve the efficiency of approximately 50% when compared with the whole image processing method. In the experiment, there are totally 2,849 logic registers in the proposed hardware design and the real-time processing in 150MHz clock rate (320 x 240 @30 fps) is achieved.