Zhi-Song Liu

Title: Zhi-Song Liu: Innovator in Neural Network Technology

Introduction

Zhi-Song Liu is a prominent inventor based in Hong Kong, CN. He has made significant contributions to the field of image processing through his innovative patent. His work focuses on enhancing low-light images using advanced neural network techniques.

Latest Patents

Zhi-Song Liu holds a patent for a "Neural-network for raw low-light image enhancement." This invention involves a neural network that receives mono-color pixels in a Bayer pattern from an image sensor. It interpolates these pixels to generate full-color RGB pixels while also enhancing the image for better detail. The design includes a Back-Projection (BP) layer added to each contracting layer in a U-net convolution neural network. This structure increases feature depth as pixels are downsampled. Additionally, a BP channel shrink layer is incorporated into each expansion layer to upsample and increase pixel resolution. The convolution layers in the BP layer lighten and darken images to generate an error that is then lightened and added to a weighted lightened input. The neural network learns the optimal weightings to correct for noise and error in the lightening process within these BP layers. Noise is further reduced by repeating a convolution and leaky Rectified Linear Unit (lrelu) in series for the first convolution in the BP layer that performs lightening.

Career Highlights

Zhi-Song Liu is affiliated with the Hong Kong Applied Science and Technology Research Institute. His work at this institution has allowed him to explore and develop cutting-edge technologies in image processing and neural networks.

Collaborations

Zhi-Song Liu has collaborated with notable colleagues, including Yan Huo and Li-Wen Wang. Their combined expertise has contributed to advancements in their respective fields.

Conclusion

Zhi-Song Liu is a key figure in the development of neural network technologies for image enhancement. His innovative patent demonstrates his commitment to improving image processing techniques, particularly in low-light conditions.