Research on resistor array non-uniformity correction with neural network

In order to make the imagery respond just as the real-world scenes do, nonuniformity correction of resistor array is necessary. Most of all, this paper proposed a new neural network algorithm for nonuniformity correction. It has simple neural network model, at the same time, not only can well compensated nonuniformity, also can fit the characteristic curves. In addition, we adopted piecewise linear method to approach the correspondence curve between the actual input voltage and the theoretical input voltage for the first time. This approach greatly reduces the storage space and makes the algorithm easily realized in hardware. Theoretical analysis of the improved neural network algorithm, the algorithm´s implementation steps, piecewise linear approach of the corresponding curve, simulation results and considerations are given in the paper.