Deep image denoising network (DnCNN)Credit: Harbin Institute of Technology
Research at HIT’s School of Computer Science and Technology (SCST) ranges from computing theory, artificial intelligence, and software engineering, to cyber security, bioinformatics and natural language processing. Since its founding in 1956, the school’s work has contributed to the aerospace industry, national security and economy, and social development.
The inputting of thousands of Chinese characters needed for computers to process language presents a challenge. SCST’s sentence-level pinyin input exploits contextual information for intelligent spelling-character conversion, becoming the most popular Chinese input tool. SCST researchers have developed a comprehensive Chinese Language Technology Platform (LTP), providing word segmentation and syntactic analysis for natural language processing. Used by hundreds of corporations and universities, LTP has won many awards.
SCST focuses on computer image enhancement and compression, visual understanding, and multimedia analytics. Some of its image/video compression techniques have become Chinese national standards. SCST’s image denoising method, based on convolutional neural networks has gained more than 1,700 citations in the past three years, and has been included in the Matlab R2017b Image Processing Toolbox and Deep Learning Toolbox.