Finding minimum bounding box of item from an image

Hello I need a C++ 64 bit program written in Visual Studio 2015 build with static .ddls that can load an image, find the minimum bounding box of each item from an image and save the resulting image with drawn bounding box(es) as bmp file. The bounding box should have the edges parallel to the axis. The algorithm should be very fast (for images attached the scanning and decision should take 5 - 15 ms). The user interface of the program must allow : - selecting an image (by using a file dialog) - starting the processing ( by pressing a start button ) - saving the result ( by using a file dialog ) Regardless of the number of the items, there should be maximum 16 bounding boxes. If there are more than 16 items, there should be bounding boxes that must contain more than one item. If you think that some optimizations can be made for this case, contact me. There are several pictures attached as examples (gray only images) the pixels that are completely black are not part of an item. Before accepting your bid , I want to know the details about implementing the project. Thank you in advance