Analysis of Raw Sensor Data with Applications in Image Processing and Compression
- Author(s): Lee, Yeejin
- Advisor(s): Nguyen, Truong
- et al.
In the last few years, there has been a drastic improvement in the development of sensor technology that increases spatial resolution, dynamic range, and low-light sensitivity of digital cameras. Although the image processing techniques in a digital camera processing pipeline have been well studied, they face many difficulties in processing raw data of high resolution as well as raw data captured in the low light environment.
The key to maximizing the quality of the final output images is to understand the captured raw sensor data in the proper context of a camera processing pipeline. In this dissertation, we analyze the image acquisition model and develop new image processing techniques that leverage the acquisition model. This work advances computer vision and data communication by focusing on the role that the camera processing pipeline plays. Particularly, we discuss image enhancement in low-light condition and image compression problems in the context of the image pipeline.
To the best of our knowledge, this is the first study connecting image enhancement and compression algorithms to the context of an actual image acquisition process. Most prior image enhancement and image compression for raw sensor data used images already processed by camera processing pipelines for experimental verification while we verify the proposed image processing technique using actual sensor data.