A correlation map shows the correlation between an image region and another image computed at the neighbourhood of each point of the second image. Most cross-correlation techniques fail when the viewpoint rotation grows over a certain value, usually around 20 degrees. The approach presented here aims at resolving this problem and at the same time provide an estimate of the actual rotation. This paper presents a method to obtain a correlation map robust to rotation together with an orientation map that gives an estimate of the rotation between the region and the image at each point. We call this a Rotation Correlation Map (RCM) and is built using texture and intensity information consecutively. The gradient histogram (texture distribution) of the region is used as a fast strategy to locate possible points of high correlation and an estimate of their orientation with respect to the region that is compared. Among the candidates with high histogram similarity, Normalised Cross Correlation (NCC) (using intensity) is computed using the orientation estimated in the previous step. Results show the accuracy in correlation and orientation and the overall higher performance when compared to similar maps