Objective:
This project was developed as part of a smart electricity billing application based on research work by Elrefaei et al., [1]. It has the potential to save time, prevent human errors during electricity bill generation.
Four units namely- Volt, Ampere, Kilowatt, Kilowatt-Hour, were extracted in real time from digital electricity meter videos using OpenCV and Python. Please see the report in this repository for a detailed documentation.
Method:
-
First videos were pre-processed to extract the digital display area using color filtering and contour detection.
-
After extracting screen, template matching was tried to extract key metrics.
-
The results were improvised by using feature matching.
- Various algorithms like SIFT and ORB were tried for feature extraction.
- Brute force and Flann based techniques were tried for feature matching.
- Then ratio test was used to extract good points.
-
Identified units were displayed on screen
Following images show the result of ORB with Flann technique.
Earlier process to take the meter readings and generate bills involved a company official to make home visit, note down the reading then a bill is generated at the office and couriered to the home.
The project had been developed as a partial fulfillment of the Practice School 1 course at BITS Pilani. The project has been developed under the guidance of Senior Scientist at BISAG, Gandhinagar. The software was accepted by Gujarat Power R&D Department for further development. I received the Innovation Challenge Award and cash prize by Dept. of Science and Technology, Govt. of Gujarat for high impact results.
Reference [1] : Elrefaei, L. A., Bajaber, A., Natheir, S., AbuSanab, N., & Bazi, M. (2015, November). Automatic electricity meter reading based on image processing. In 2015 IEEE Jordan Conference on Applied Electrical Engineering and Computing Technologies (AEECT) (pp. 1-5). IEEE. Link