fit rectangles into a 2D shape within an image using either Matlab or Python -- 2

Job Description:

The task at hand requires some programming skills and a basic understanding of math. There are some images, each depicting a shape such as a handwritten digit, alphabet, etc. The physical dimensions of the shape are approximately 3 by 3 micrometers. The goal is to fit smaller rectangles (with side lengths between 20 nm and 500 nm) into the object shape (general 2D shapes) in an optimized manner. The rectangles should be either vertical or horizontal and their arrangement, size, and number should be optimized so that the sum of the rectangles closely resembles the original image.

For example, the code reads an image [login to view URL] (see attachment) and assembles rectangles, similar to the sketch in [login to view URL] (see attachment). Then, returns the center positions, length, and width of each rectangles.

The shape of the objective is general, it could be images of digits, alphabets or simple objects like ones in the attached files.

The script should be able to read an image and return the optimized length (a) and width (b) of the rectangles, their central position, and an indicator of how well the shape is approximated by the rectangles. Gaussian process regression (Bayesian optimization) is a suitable method to solve this problem.

The code should be well-commented for easy understanding. After the code is written and tested by you, I will further test it to verify.

Deliverable is a working code (python or Matlab) that does what it supposed to do.

Skills: Python, Matlab and Mathematica, Algorithm, Machine Learning (ML), Image Processing

Project ID: #35913601