Local Coupled Extreme Learning Machine Based Image Watermarking using DCT in YCbCr Space
Ashok Kumar Yadav, Rajesh R. Mehta · 2019
Recently developed single layer feed forward neural network model (SLFN): local coupled extreme learning machine (LC-ELM) based color image watermarking in DCT domain is designed in the proposed scheme. LC-ELM is used to handle non-linear regression analysis by considering the image watermarking problem as a regression problem. Firstly, the RGB image is transformed into YCbCr to obtain high de-correlation in YCbCr space to accomplish good imperceptibility and robustness. After transformation, DCT is used on the selected blocks depending on the randomness measurement of Y component. The features extracted from the selected blocks helps in making image dataset and embedding the watermark using trained LC-ELM. The experimental results performed on color images confirmed the generalization performance of LC-ELM in addition to imperceptibility and robustness.