Automatic face recognition is a major ramification in the field of Artificial Intelligence. Local binary patterns are designed, such as LBP, GLBP, VLBP, and more extended variants were introduced, but many challenges, like face recognition in various illumination conditions, pixel intensity variation, noisy threshold function, and lower recognition rate etc., have not been addressed. In this paper, we propose a novel feature extraction approach for illumination-insensitive facial recognition and object recognition, as well as under the various illumination conditions named LVIP (Local Vertical Index-based Patterns). This pattern helps to extract features under the varying lighting conditions. Eventually, we also proposed a new similarity measure method based on the feature integration of rudimentary measures to retrieve similar images. The proposed method was investigated on datasets such as Multi–PIE, VGG Face2, IJB, ExDark, and proved to perform better results in terms of precision, accuracy, and recognition rate.