This project aimed to automate the reading of analogue gauges using image processing techniques, with a specific focus on mitigating the negative effects of glare and shadow on the readings. Despite the constraints of an 8-week internship, the primary objective was to conduct data collection and preprocessing to address these challenges. This README provides an overview of the project, detailing the methodologies employed, challenges faced, and future directions.
The project revolved around leveraging computer vision methods to analyze images of analogue gauges, extract pertinent information such as needle position, and derive accurate readings while compensating for glare and shadow effects. The provided code demonstrates the initial steps of the process, including image loading, preprocessing techniques, and visualization of intermediate results.
The Python script utilizes OpenCV and NumPy libraries for image processing tasks. Key steps performed include:
- Image Loading: The script loads the input image using OpenCV.
- Grayscale Conversion: Conversion to grayscale simplifies subsequent processing steps and is less susceptible to glare and shadow.
- Adaptive Thresholding: Adaptive thresholding enhances local contrast, aiding in the segmentation of the gauge from the background.
- Binary Thresholding: Further refinement of the segmentation is achieved through binary thresholding.
- Morphological Operations: Erosion and dilation operations are applied to remove noise and restore the original object size, mitigating the impact of glare and shadow.
- Smoothing: Median and Gaussian filtering techniques are employed to further reduce noise and enhance image quality, contributing to more accurate needle detection.
Several challenges were encountered during the internship:
- Glare and Shadow: Mitigating the adverse effects of glare and shadow on gauge readings required careful preprocessing and algorithm selection.
- Limited Dataset: The availability of diverse gauge images with varying lighting conditions was limited, hindering the robustness of the solution.
- Time Constraints: The 8-week timeframe restricted the depth of exploration and implementation of advanced algorithms for glare and shadow compensation.
Despite the constraints, the project lays the groundwork for future advancements:
- Dataset Expansion: Collection of a more extensive and diverse dataset encompassing various lighting conditions to improve model generalization.
- Real-Time Implementation: Adapting the solution for real-time processing in practical applications such as industrial automation and monitoring systems.
- Integration with Machine Learning: Exploring the integration of machine learning models to learn and adapt to different glare and shadow patterns.
The internship focused on addressing the challenges of glare and shadow in analogue gauge reading automation through data collection and preprocessing. While the full objectives were not achieved within the internship duration, significant progress was made. The provided code and documentation serve as a foundation for future research and development in this domain, with the potential to enhance accuracy and robustness in gauge reading automation systems.