Jinnan Hu

Title: Jinnan Hu: Innovator in Image Denoising Technology

Introduction

Jinnan Hu is a prominent inventor based in Hangzhou, China. He has made significant contributions to the field of image processing, particularly in the area of image denoising. His innovative approach addresses common challenges faced in the processing of noisy images.

Latest Patents

Jinnan Hu holds a patent for an "Image denoising method and apparatus based on wavelet high-frequency channel synthesis." This patent discloses a method that expands image data into multiple frequency-domain channels. It utilizes a noise-sort algorithm to group channels into 'less-noise' and 'more-noise' categories. The invention combines a denoising submodule with a synthesis submodule based on style transfer to create a generative network. A discriminative network is also established to impose constraints on the global loss function. The iterative training of the GAN model allows for the extraction of denoised image data through wavelet inverse transformation. This algorithm effectively mitigates issues of blurring and loss of details that are often introduced by traditional filtering or CNN-based deep learning methods. It is particularly beneficial for handling noise-overwhelmed or high-dimensional image data.

Career Highlights

Jinnan Hu is associated with Zhejiang Lab, where he continues to advance his research and development in image processing technologies. His work has garnered attention for its innovative solutions to complex problems in the field.

Collaborations

Jinnan collaborates with notable colleagues, including Jingsong Li and Peijun Hu, who contribute to his research endeavors. Their combined expertise enhances the development of cutting-edge technologies in image denoising.

Conclusion

Jinnan Hu's contributions to image denoising technology exemplify the impact of innovative thinking in solving real-world problems. His patent and ongoing work at Zhejiang Lab position him as a key figure in the advancement of image processing methodologies.