ITSnoise: An Image Quality Dataset With Sensor Noise
Joel M. Dumke, Margaret Pinson · 2022
This technical memorandum provides details for the image quality subjective experiment ITSnoise. This experiment includes 24 scenes, each digitally photographed using 12 different Image Capture Methods (ICMs) for a total of 288 images. The ICMs were designed to reflect the way images are captured in the most common public safety applications and also to generate a range of different sensor noise levels in the resulting images, from very low to very high. This dataset was intended to provide training and testing data for potential no-reference sensor noise metrics that could automatically predict the impact on perceived quality of sensor noise within a given image. This dataset is freely available for research and development purposes.