Plug-and-play high-frequency feature enhancement for plant image super-resolution

Area of Science:

• Plant Science
• Computer Vision
• Digital Agriculture

Background:

• High-resolution plant imagery is crucial for phenotyping, disease monitoring, and precision agriculture.
• Existing super-resolution (SR) methods struggle to recover essential fine structural details in low-quality plant images due to sensor limitations and environmental noise.

Purpose of the Study:

• To introduce a novel plug-and-play high-frequency feature enhancement (HF-FE) module for improving plant image super-resolution.
• To enhance the reconstruction of subtle plant features like leaf venation and lesion boundaries.

Main Methods:

• Developed a HF-FE module designed for seamless integration into existing SR architectures.
• Evaluated the module's performance on diverse plant datasets including oil palm, aquatic plants (AqUAVPlant), and crop disease imagery (Plant Pathology 2020).

Main Results:

• Models with the HF-FE module demonstrated consistent improvements over state-of-the-art baselines across all tested datasets.
• Significant gains in Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index Measure (SSIM) were observed.
• Visual analysis confirmed enhanced clarity of fine structural features in plant images.

Conclusions:

• The HF-FE module offers an effective and flexible strategy for enhancing plant image SR.
• Improved image fidelity supports more accurate plant visualization and analysis.
• This advancement contributes to intelligent plant sensing, digital agriculture, and sustainable crop management.