A Ghaffari - Electrical Engineering Department Sharif University of Technology, Tehran, Iran
E. Fatemizadeh - Electrical Engineering Department Sharif University of Technology, Tehran, Iran

Similarity measure is an important key in image registration.Most traditional intensity-based similarity measures (e.g.,SSD, CC, MI, and CR) assume stationary image and pixel bypixel independence. Hence, perfect image registration cannotbe achieved especially in presence of spatially-varying intensitydistortions and outlier objects that appear in one imagebut not in the other. Here, we suppose that non stationaryintensity distortion (such as Bias field or Outlier) has sparserepresentation in transformation domain. Based on this assumption,the zero norm (ℓ0) of the residual image betweentwo registered images in transform domain is introduced as anew similarity measure in presence of non-stationary intensity.In this paper we replace ℓ0 norm with ℓ1 norm which is apopular sparseness measure. This measure produces accurateregistration results in compare to other similarity measuresuch as SSD, MI and Residual Complexity RC.

image registration, Bias field, nonstation ary intensity distortion, outlier, sparse representation, sparseness ness