Eccentricity compensator for wide-angle fovea vision sensor

This paper aims at acquiring robust feature for rotation-, scale-, and translation-invariant image matching from a space-variant image by a fovea sensor. A proposed model of eccentric compensator corrects deformation in a log-polar image when the fovea sensor is not centered at a target image, that is, eccentricity exists. An image simulator in discrete space implements this model by its geometrical formulation. This paper also proposes Unreliable Feature Omission (UFO) using Discrete Wavelet Transform. UFO reduces local high frequency noise appeared in the space-variant image when the eccentricity changes. It discards coefficients when they are regarded as unreliable, based on digitized errors of the input image by the fovea sensor. The first simulation estimates the compensator by comparing with other polar images. This result shows the compensator works well and its root mean square error (RMSE) changes only by up to 2.54 [%], in condition of the eccentricity within 34.08 [°]. The second simulation shows UFO works well for the log-polar image remapped by the eccentricity compensator, when white Gaussian noise (WGN) is added. The result by Daubechies (7, 9) biorthogonal wavelet shows UFO reduces the RMSE by up to 0.40 [%] even if the WGN is not added, when the eccentricity is within 34.08 [°].