The proposed method focuses on the Image forgery detection using Hybrid Approach. The proposed method is divided into three phases. In first phase apply K-Means clustering for R, G, B channel of an input image and then the Gaussian filter is applied to blur an input image. Then entropy feature is extracted from both color and gray scale of an input image and store it in matrix form. In second phase the extracted features are matched and mark the matched blocks with blue and red color. In third phase count continuous positions of matched blocks, if it is greater than 10 pixels then displayed as forged or else non forged image. From the experimental result it is shown that the hybrid approach is detecting the forgery in an image with high accuracy.