Exposure Response and Noise in a Digital Imaging System

Overview

Pictures represent scenes by recording different values of brightness and color at different locations. A typical digital camera has a sensor with millions of pixels. Each pixel has stored electric charge that leaks out through a photodiode when light falls onto it. This charge is converted to a voltage and a binary number, which represents the camera exposure. This should correspond to the amount of light energy that hit the pixel (radiant exposure). Variations in camera exposure for the same radiant exposure represent noise. The range of radiant exposures that can be recorded represents the dynamic range. Besides the scene's luminance (brightness), exposure and noise depend on the camera settings of shutter time value, aperture value, sensitivity (ISO), and resolution (megapixels). The following experiments used a Canon digital camera and a software called GetRGB to extract the camera exposure values for every individual pixel in the picture. The calculated mean and standard deviation for all pixels reflected the overall exposure and the noise respectively. The first experiment’s goal was to test the proportionality between the camera exposure and radiant exposure. The second experiment’s goal was to test the principle of reciprocity, in which the same camera exposure should be achieved with different combinations of time and aperture values as long as the same radiant exposure is maintained. The third experiment measured noise as ISO was increased. The fourth experiment’s goal was to see how noise is affected by changes in resolution. The fifth experiment studied noise versus time value. Finally, the noise and dynamic range of this same basic digital camera were compared to a more advanced camera. The following conclusions were found: Even a basic digital camera proved to have a more linear exposure response and better reciprocity than typical film. Then, as expected, the noise increased with ISO due to a higher amplification of a smaller voltage. Also the noise increased with exposure time because of the dark current. As I hypothesized, a decrease in the number of pixels resulted in less noise due to the merging of pixels which collects more light energy with fractionally less variations; this contradicts the common misconception that more pixels are always better. Finally, this project provides quantitative information on how optimize the ISO and time values to reduce noise and improve image quality, and provides a quantitative way to assess and compare digital cameras.

Competition history

• AJAS 2018 Category not listed

Related projects

Closest projects by meaning, across every fair and year in the corpus.

Browse more like this

• AJAS None projects
• Every AJAS 2018 project