Yes, the approach of image segmentation depends on type of input image and complexity associated with the ROI and back ground objects. In disctinct RGB image, you can use color based segmentation.
For other you can create a mask and apply the mask on input image, approaches are thresholding (multiple ways, global, local), distance transform, very popular Otshu's methd (Thresholding), watershed, edge detection etc and followed by morphological operation (if required)
Some cases you get the very good results, if you apply contrast enahancement (Histogram Equalization) before apply the segmnetaion method, there after only proceed to subsequents work.
Highly Recomended: Please refer Gonalez Book of Image Segmnetaion Chapter to get the foundation.
Hope it Helps!