Performance of contrast-limited AHE in preprocessing of face recognition with training image under various lighting conditions

Face recognition performance is greatly influenced by the lighting conditions on the face images. Generally, to solve the problem caused by this lighting factor, the illumination normalization process is carried out at the preprocessing stage through contrast adjustment of face images. One of the contrast adjustment techniques is Contrast-limited AHE. In many cases, this technique performs well for normalizing the illumination, thereby improving general object recognition performance. In this research, an experiment was conducted to determine the effect of contrast adjustment using Contrast-limited AHE on face recognition in each lighting condition in the image. So later, it can be seen whether the technique is useful in all image lighting conditions or not. The Robust Regression method is used for the classification process in face recognition, which has an excellent performance in recognizing faces influenced by lighting factors. Experiments use facial images in the AR Face Database related to lighting factors, with low, medium, high, and very high lighting conditions. The testing process is carried out by comparing face recognition performance using the Contrast-limited AHE technique in the preprocessing phase and face recognition without a contrast adjustment process at each lighting condition. Experimental results in the low, medium, and high lighting conditions show that using the Contrast-limited AHE technique in preprocessing results in face recognition performance much better than face recognition without preprocessing. However, in very high (extreme) lighting conditions, the use of the Contrast-limited AHE technique in preprocessing has a less significant effect on face recognition performance, which results in an average accuracy of 95.33%, only 0.67% higher than face recognition without preprocessing.