A hardware computing system and method for image up-sampling based on the Canny algorithm
By developing an image upsampling hardware computing system based on the Canny algorithm, the problems of high hardware resource consumption and slow computing time are solved, and the improvement of image edges and details is achieved. This system is suitable for image processing and video communication.
Patent Information
- Application Number
- CN202211357168.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-11-01
- Publication Date
- 2026-07-24
- Estimated Expiration
- 2042-11-01
AI Technical Summary
In existing technologies, image upsampling operations consume a lot of hardware resources and have a slow computation time, making them impossible to implement simply through integrated circuit design. Furthermore, traditional methods result in severe jagged edges on the images, while neural network methods consume a lot of computational resources and cannot be processed in real time.
A hardware computing system for image upsampling based on the Canny algorithm is adopted, including an IP control unit, an SRAM storage module, an image boundary and core interpolation control module, linear and bicubic interpolation units, a synthetic gradient calculation unit, and an edge detection and replacement unit. By using line-by-line calculation and edge optimization methods, the hardware resource consumption is reduced and the computing speed is improved.
It achieves image edge and detail improvement with low hardware resource consumption, can process images in real time, reduces storage resource consumption and improves computing speed, and is suitable for integrated circuit design, image processing and video communication fields.
Smart Images

Figure CN115526903B_ABST
Abstract
Citation Information
Patent Citations
An edge-guided adaptive image interpolation method and its VLSI implementation device
CN103500435B
Method for extracting human ear image edge combining multiple methods
CN101256625A
Single image based edge-preserved interpolation super-solution calculation method
CN107194877A