Generation of Netlist from a Hand drawn Circuit through Image Processing and Machine Learning

Abstract

Circuit diagrams are used to depict electronic or electrical circuits graphically. It is simple for everyone to put their thoughts on paper. However, in order to conduct simulations in the different available tools, the circuit model needs be in digital form. This project presents several image processing and machine learning approaches for the conversion of hand-drawn circuits to netlists. Rather than training the dataset for all components, a technique based on the length ratios of a few of lines was employed to identify elements such as a voltage source, ground, and capacitor. Various image processing techniques are used to eliminate noise and prepare pictures for further processing. HOG feature extraction is utilized throughout the training and segmentation stages to detect resistor, diode, and inductor components. The final stage is to construct a netlist from the detected elements, wires, and their locations, as well as the identified nodes.