Tongue diagnosis is one of noninvasive methods to diagnose the condition of a patient’s internal
organs in Traditional Chinese Medicine (TCM). Since this way is noninvasive, it encourages self-diagnosis
at home. One of the essential aspects of this diagnosis is the tongue color which is prone to being influenced
by the lighting environment and different sensor sensitivity. It brings ambiguity and problems to
get a consistent diagnosis result in self-diagnosis. In facing this problem, many researchers have found
great solutions to make a reliable automated tongue diagnosis (ATD) system which potentially can solve
the problem for a smartphone. However, the system can still be improved to minimize the error which is
caused by the different sensor sensitivity and unoptimized parameters. For improving it, this paper suggests
an alternative framework for tongue color classification system (TCCS) as the part of ATD system in getting
more consistent color before going to the disease prediction in ATD system. The improvisation is done by
implementing the kernel partial least square regression (K-PLSR) optimized algorithm in color correction
of framework and other reliable algorithms. After doing an experiment by implementing this framework in
several smartphones, this framework can show an improvement and can be used in more than one sensor
and environment. As a result, it gets a more consistent and objective tongue color classification result.