In color images, out-of-focus problems often occur when different wavelengths of rays are focused at different positions in the focal plane. This occurs because of lenses that have different refractive indices for different wavelengths of light. These color images in turn become blurred, and noticeable colored edges appear around objects. These misaligned edges thus degrade the overall quality of the images. In this study, we propose a restoration algorithm for misaligned edges in color images. This algorithm is based on the assumption that the gradients of color channels are highly correlated such that all edges spatially overlap in the same manner as the desired gradient. The constraint term in least squares optimization is proposed to align edges to match the desired gradient based on the transformation theory of gradient profile sharpness. The proposed constraint term adaptively uses the gradients of the channels with different weights to estimate sharp edges. We also design a new measurement to compute the energy of aligned edges in a color image. The proposed algorithm can be applied to images captured by various sensors in different environments. Experimental results show that the proposed algorithm performs effectively when estimating high-quality color images.
All Science Journal Classification (ASJC) codes
- Signal Processing
- Applied Mathematics