This study aims to investigate the different local thresholding methods on various regions of noise images, which could be used for image binarization of optical coherence tomography images. In the methods one hundred 8-bit images of noise, 1000x1000 pixel in size, is generated using ImageJ/FIJI program. Images processed with four different auto local threshold method in ImageJ/FIJI program as Niblack, mean, median and midgrey, in order to perform binarization. Twenty-five different region of interest, 100x100 pixel in size, from different region in an image analyzed for area percentage (AP) measurement. Normality tests were performed via Saphiro Wilk Normality test, and Student’s t test and one-way ANOVA were used to assess the continuous variables, and Bonferroni test for post hoc analysis, utilizing the IBM SPSS Statistics, Version 28.0 for the statistical analysis. In the results of this study mean AP for Niblack method was 42.08 ± 0.32%, for mean method was 50.00 ± 0.32%, for median method was 49.28 ± 0.16%, and for midgrey method was 49.63 ± 2.09%. One-way ANOVA analysis shows all of the different subgroups of Niblack and mean, Niblack and median, Niblack and midgrey, mean and median, mean and midgrey, and median and midgrey measurements are significantly different from each other (p < 0.001 for all of the subgroup post hoc analysis). In conclusion our study examined 100 noise images across 25 regions using four auto local threshold methods (Niblack, mean, median, and midgrey). Analyses indicated that Niblack having the lowest mean and there is significant difference between all of the methods; researchers using auto local threshold methods in OCT image processing should select methods aligned with data properties, warranting further exploration of these methods' impact on diverse OCT image, esspecially taking into account the effect of the noise.