Research

We are conducting researches on multi-media signal processing and image processing. Our Lab. researches are for practical uses or researches and developments near for real manufactures such as algorithm researches and developments, implementations on DSP, FPGA and GPU for some kinds of image processings and so on. The details are shown as follows:

We have been studying on super-resolution image reconstruction technique, which has become a hot topic in high-definition televisions. Some kind of edge-cut technologies such as Total Variation regularization method, learning-based method, multi-frame method have been being studied, and our goal is to implement them on next-generation high-definition television in the future.

Super-resolution image reconstruction, which enables us to enlarge high quality images, is used in medical images to aid in the diagnosis of medical conditions. We have been developing an application that automatically measures Joint Space Distances by focusing on the disease of rheumatoid arthritis.

Super-resolution processing, which can enlarge license plate images captured by security cameras and in-vehicle cameras more clearly than normal image enlargement, is used to help read characters and programmatically recognize characters.

We have been studying on noise reduction techniques which are used for total variation regularization method.

Photos, which we have been taken, have degradation suffer from blur. To solve these problems, we have been utilizing Total Variation regularization, Shock Filter and learning-based methods to restore the blur.

Our research focuses on practical uses of image recognition technology to recognize and detect objects and features such as characters and faces from still images and moving pictures.