Farhan A. Baqai

Analysis and extensions of the Frankle-McCann Retinex algorithm

We present our analysis of the Frankle-McCann Retinex algorithm and propose extensions that result in a better rendition of the image. These extensions suppress artifacts and reduce the number of iterations required to get to the final image. In addition, our investigation suggests that the Retinex algorithm can be viewed as a process of estimating the per-pixel gains necessary to achieve dynamic range compression in the original image. Experimental results are provided that demonstrate the efficacy of this approach as compared to using Retinex to generate the final image.

Color constancy using fractals

We combine fractal decompression and the Retinex algorithm do devise a new color constancy method. We show that by using this approach, we can achieve color constancy and image compression simultaneously. Experimental results are included that show that this approach is quite promising.

Fast linear method of illumination classification

We present a simple method for estimating the scene illuminant for images obtained by a Digital Still Camera (DSC). The proposed method utilizes basis vectors obtained from known memory color reflectance to identify the memory color objects in the image. Once the memory color pixels are identified, we use the ratios of the red/green and blue/green to determine the most likely illuminant in the image. The critical part of the method is to estimate the smallest set of basis vectors that closely represent the memory color reflectances. Basis vectors obtained from both Principal Component Analysis (PCA) and Independent Component Analysis (ICA) are used. We will show that only two ICA basis vectors are needed to get an acceptable estimate.