In order to effectively simulate a digital camera, knowledge is required of principal image sensor characteristics. Important factors to include in this analysis are the following: dark current, dark signal non-uniformity (DSNU), read noise, spectral response function and photo response non-uniformity (PRNU). We set out to characterize and simulate a specific camera, the Nikon D70, borrowed from the Stanford Electrical Engineering department.
Furthermore, we utilized these characterization values, as well as the Imaging Systems Evaluation Tool (ISET), developed by Professor Brian Wandell, to simulate the Nikon D70. Using the Macbeth Color Checker under tungsten lighting, we set out to compare the actual performance data obtained from the Nikon D70 to a simulated ISET version based on our characterization results.
The Nikon D70 is a digital Single Lens Reflex (SLR) camera first introduced in January 2004. With approximately 6.1 million CCD pixels, the Nikon D70 is particularly useful for analysis, because it allows access to the ‘raw’ pixel data in addition to the typical JPEG output. Thus, we can freely analyze this raw data using MATLAB imaging processing software and characterize key parameters of the camera.
The Nikon D70 saves the ‘raw’ data in a .NEF file which holds the four RGBG channels, each 1007 by 1520 pixels. The Nikon D70 compresses its 12-bit data before the .NEF format; hence, pixel values in the ‘raw’ data range only from 0 to 2047.
The Stanford EE department very recently acquired the Nikon D70 digital camera. Hence, this particular camera has never before been characterized and simulated in the ISET simulator. However, previous students in the EE 362 course have simulated other models using the ISET simulator. Because the Nikon D70 is a commercially available camera, obviously the manufacturers have characterized this camera model before releasing it to the public.