This research Paper represents a recent use of the extended local binary pattern for face recognition. Extended Local Binary Pattern (ELBP) Technique is more accurate and describes the texture and shape of a digital image by using of 3*3 & 5*5 matrices we have to compare the performance of both matrices so that how we recognize the image. Variance help to measure continuous output where the quantization is needed. By dividing an image into several small region from which the feature are extracted. If match is found then image face is recognized otherwise if match does not found then image face is not recognized. If we saw at the mirror we can see that our face has different type of human expression. These are the peak and valley that make up the different facial features.