There might be several ways to do this. A straight forward way I could think of is to use the template matching algorithm in the Computer Vision System Toolbox with the vision.TemplateMatcher system object. Here is an example where we try to find Location of a Specific Chip in Image of Electronic Board. You have a reference image and try to find similar objects in another image.
Based on what the actual object is that you are trying to locate, there are so many other ways ranging from simple image processing to complex neural network based solutions that could be used to solve this problem.
