NHK is accelerating development of ‘full-featured’ 8K broadcasting equipment. We have developed an 8K 120 Hz single-chip camera system by adopting 2:1 inter-line scanning. We performed an image capturing experiment, and confirmed that the resolution in the still area is the same as that in 60 Hz scanning. The signal-to-noise ratio of 120 Hz inter-line videos was improved by adaptive noise reduction technique.
In basketball video 3D analysis, it is an important issue to estimate camera pose and line detection. In this study, we focus on a dedicated court wish few lines and estimate a camera posture determination based on straight line detection and its correspondence between frames.
In this paper, we propose a method for estimating the size of an object from a single image taken with a monocular camera. We realize it by using a convolution neural network (CNN) for depth estimation and information of a plane and a reference object standing upright on the plane.
In this paper, we propose a novel localization method for autonomous driving cars and mobile robots using 3D point cloud. Our method is based on similarity search between environment map data and pose-transformed scan data. Our method improved mean accuracy by 8.35 times or more compared with conventional methods.
Object segmentation and Classification from Point cloud LiDAR are challenging due to its sparseness and lacking of detailed information. In this work, data association between point cloud and RGB image is used to automatically segment objects, which consist of car, cyclist, and pedestrian, for effective making point cloud dataset.
Derivation of equation aimed for distance measurement using omni-directional camera is studied. By using ideal images created by computers graphics correctness of the proposed equation is confirmed.
We conducted experiments on the effectiveness of an automated audio description system for visually impaired people and evaluated the degree of understanding of a game compared with live TV and radio commentary. The results showed that the automated audio descriptions and live radio commentary had almost the same mean opinion score.
Speech synthesis is expected to be utilized for distribution of emergency disaster information via Set-Top Box. Speech synthesis technology supports distributing information in visual and audio for elderly. Here we report analysis on the results of speech synthesis quality for elderly.