We performed a comparison of the results of recognition of fragmented outline images in the presence of noise and without it. Contour images and visual noise were synthesized form Gabor elements. Distances between the Gabor elements in contour and in noise were changed and image sizes varied too. The percentage of correct responses was not dependent on the size of the stimuli, but differed for the same objects in the noise and without noise. The percentage of recognition was higher for images with lots of turns in case without noise. The recognition in noise was better for contour images having lines with slightly varying curvature. Identification in noise depended on the ration of the distances between the elements in noise and in contour.