Visual enhancement of underwater image by white-balanced EMD

In this paper, an algorithm has been proposed based on Empirical Mode Decomposition (EMD) with a white balanced input to enhance the visual quality of the under-water images. Generally, the underwater images are whisked by blurring effect, scattering effect etc. because of poor lighting condition. EMD is a signal decompose algorithm which is particularly useful for non-stationary and non-linear signals. First of all, the image is processed through the Gray World technique which is a white balance approach to enhance the contrast of the image and to remove the unwanted color cast in the image. Then, each R, G and B channel of the resultant image is decomposed into its Intrinsic Mode Functions(IMFs). Final enhanced image has been constructed by combining the IMFs of each channel with different optimised weights. The EMD algorithm is implemented on the resultant image of White balanced process to restore the color. Our experimental results enhance the contrast of the image by reducing noise as well as artifacts in the image. To show the quantitative enhanced result, Gray Level Cooccurrence Matrix(GLCM), Peak Signal to Noise Ratio(PSNR) and Mean Square Error(MSE) are calculated and compared with different conventional enhancement methods. The proposed method results in superior enhanced image with increased visual quality compared to conventional methods.