Split and Merge Multi-scale Retinex Enhancement of Magnetic Resonance Medical Images

Abstract

Image contrast enhancement and highlighting the prominent details based on edge preserving are the fundamental requirement of medical image enhancement research. This paper presents the Multi-Scale Retinex (MSR) enhancement of Magnetic Resonance (MR) medical images using split and merge technique. The main limitations of the retinex based image enhancement schemes is computational complexity, halo artifacts, gray world violation, over enhancement etc. There exist various extensions for retinex methods developed by over a dozens of researchers, but most of the techniques are computationally complex. This is due to the fact that the image details are improved at the cost of increased computational complexity. The proposed method is efficient in computation, since the original input image is splitted into a size of 8 × 8 and then gray level Fast Fourier Transform (FFT) based retinex algorithm is exploited to improve the quality of sub-image. Reconstructed image is produced by merging each of the enhanced versions to the original image resolution. This scheme is validated for MR medical images and the parameters of the retinex method is adjusted to improve the contrast, details and overall quality image. Experimental results presented confirms that the proposed method outperforms compared to existing methods.