Compensating Nonuniform OLED Pixel Brightness in a Vertical Blanking Interval by Learning TFT Characteristics

Changes in thin-film transistor (TFT) characteristics can cause nonuniform brightness in organic light-emitting diode (OLED) displays, and external compensation circuits cannot address such changes if they occur rapidly. Our system acquires a short segment of the source voltage curve of the TFT driving each pixel, during the vertical blanking interval. It then estimates the threshold voltage V
_th
and the mobility-related parameter k of that TFT, using neural networks and recurrent conditional (RC) learning to perform expectation-maximization (EM). Finally, it corrects the pixel brightness accordingly. The average errors in the estimated V
_th
and k are, respectively, 7.3 mV and 3.7 ×10
^-11
F/Vs, and we demonstrate the visual effect of compensating with these estimates.