研究目的
To propose an efficient approach for enhancing local details and color information while preserving naturalness in non-uniformly illuminated images.
研究成果
The proposed method effectively enhances image details and visibility while preserving naturalness and color information, outperforming existing methods in both qualitative and quantitative assessments, and is suitable for real-time applications due to its computational efficiency.
研究不足
The method may have computational complexity issues for very large images, and it is primarily tested on a set of 25 images, which might not cover all possible scenarios of non-uniform illumination.
1：Experimental Design and Method Selection:
The method is based on retinex theory, involving steps such as gradient calculation, low-pass filtering, illumination estimation, adaptive gamma correction, and statistical data distribution for contrast enhancement.
2：Sample Selection and Data Sources:
25 test images with various illuminations (indoor and outdoor scenes) and bright/dark regions, some captured by authors and others from literature.
3：List of Experimental Equipment and Materials:
Digital gadgets like cameras and cell phones for image capture; MATLAB software for implementation.
4：Experimental Procedures and Operational Workflow:
Steps include calculating image intensity, gradients, applying low-pass filter, estimating and modifying illumination, using adaptive gamma correction, and applying statistical transformations to form the enhanced image.
5：Data Analysis Methods:
Qualitative assessment by visual comparison and quantitative assessment using image entropy and ΔE_S00 measure for color difference.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
