Real-time images usually exhibit an amplitude in homogeneity, that poses an acceptable obstacle in image segmentation. The most common image segmentation techniques are region-based which usually depend on the homogeneity of the picture brightness in the regions of interest (ROI). However, these algorithms commonly produce inaccurate segmentation results because of the level of inhomogeneity. This suggests a brand-new region-based picture segmentation technique that can handle homogeneities in brightness during segmentation. We initially construct a local brightness clustering feature for the image intensity and develop a local clustering criterion utility to represent the image intensity in the vicinity of every point according to the model of pictures that intensity in homogeneities. A global image segmentation requirement is then produced by integrating this local clustering requirement function with regard to the neighborhood center. This criterion specifies an energy as a function of level set functions that indicate a division of the image domain and an offset field that takes into consideration the degree in homogeneous of the image in a level set formulation. Consequently, by utilizing the level set approach and reducing this energy, our approach may divide the image and approximate the bias field at the same time. The calculated bias field is able to be utilized for additive bias correction to compensate for brightness in homogeneity. Our technique is being tested on actual and fake images of different techniques, and it performs admirably even when there are homogeneities in the brightness. Experiments showed that our approach is faster, more precise, and more initialization-resistant than the recognized piecewise smooth approach. Our technique is being applied, without encouraging results, to the categorization and bias correction of images.