. 2012 Jan 5;12(1):585–611. doi: 10.3390/s120100585

Table 1.

Bit-width of each variable taking part in the calculation of colordist and brightness. The first value represents the integer part and the second value represents the fractional part. The bit-width values have been determined as the minimum values of the fractional part for which the quantization error is approximately stable. This information can be easily extracted from Figure 4.

Variable	Bits
< B_i, C_i, D_i>	[18 8]
< E_i, (S_i)⁻¹>	[8 8]
< (a_i)⁻¹, (b_i)⁻¹>	[8 10]
α_i	[28 8]
CD_i	[18 8]
α̂_i	[36 10]
${\hat{CD}}_{i}$	[26 10]