Defective pixels detection during IR FPA two-point correction
K. О. Boltar, R. V. Grachev, V. V. Poluneev
Orion Research-and-Production Association, Moscow, Russia
Detection of defective IR FPA pixels technique based on cold and hot calibrations process is presented. For cold
calibration N consequent frames are accumulated. Applying the least mean square method to N accumulated values of
every pixel, the best linear approximation is built, defining offset for this pixel and offset drift index.
Root-mean-square deviation of measured values from the approximation line evaluates the temporal noise at
frequency defined by delay between consequent frames. Defective pixels detection is accomplished by using
thresholds for drift indexes, temporal noise and offset values. Sensitivity coefficients of every pixel
calculated during hot calibration are also used for defective pixels detection. A set of defective pixel
correction methods is presented. Methods are implemented using Multicore series digital signal processor
which performs real-time processing of IR FPA frames. Spatial and statistical distributions of drift indexes
and noise versus frequency dependence are presented for infrared camera based on LWIR HgCdTe 256x256 pixels IR FPA.
PACS: 85.60.Bt
Contents