Detect IT

The goal of this project is to be able to accurately detect faults in solar PV panels. The hardware components include a DJI Phantom 3 drone equipped with a FLIR one thermal camera with a phone attached to it and Raspberry Pi 4 on which the processing takes place. The software-based process to extract these faults and identify their types is as follows. First, canny edge detection and color segmentation are applied to the taken thermal images. This process results in a binary file that highlights the location of the panels based on color & edge characteristics. However, unwanted highlights exist and must be removed using morphological operations that make sure small highlights are removed. After detecting the panels, we extract the objects or faults inside the panels that we are interested in based on some area characteristics, bounding box, centroid, and circularity. Once these objects are detected, we can identify the type of faults that each panel incurs. However, some functions in the MATLAB code are not compatible with Raspberry PI 4, so we had to design our C code that extracts this image and then sends it to our  MATLAB code. The last step was to wrap up our C and MATLAB codes in a code generator to generate one unified C code that is deployable on Raspberry Pi. 

Project Details

[photo]