High Resolution Local Structure-Constrained Image Upsampling.

Area of Science:

• Computer Vision
• Image Processing
• Display Technology

Background:

• Increasing demand for ultra-high-resolution displays necessitates improved visual perception of fine textures.
• Existing image upsampling methods often struggle to balance quality, cost, and detail preservation.

Purpose of the Study:

• To develop a fast and efficient image upsampling method for high-resolution displays.
• To enhance the recovery of fine texture details and sharp edges in upsampled images.

Main Methods:

• A novel image upsampling approach utilizing high-resolution local structure constraints.
• Segmentation of bicubic-interpolated images into sharp edge and texture areas based on average local difference.
• Separate reconstruction of areas using estimated high-resolution gradient and texture structure maps.

Main Results:

• The proposed method successfully recovers finer pixel-level texture details.
• Achieved top-level objective performance metrics compared to state-of-the-art methods.
• Demonstrated a low time cost, indicating computational efficiency.

Conclusions:

• The developed method provides a high-quality, low-cost solution for image upsampling.
• Effective reconstruction of both sharp edges and intricate textures is achieved.
• The approach is well-suited for applications requiring enhanced visual detail on modern displays.