The invention relates to a circular loop statistic characteristic-based anti-rotation image Hash method, which comprises the following steps of: normalizing the size of an image size by using a bilinear interpolation method; processing the image by using a Gaussian low-pass filter; if the image is a color image, converting the image into a YCbCr color space, and representing the color image by using a brightness component; dividing the image into a plurality of circular loops, and extracting statistical data such as an average value, variance, skewness degree and kurtosis of each circular ring as characteristics; normalizing the statistical characteristics; calculating an average value of the statistical data, taking the average value as a reference, calculating Euclidean distances between the statistical characteristics of the circular loops and reference characteristics, and connecting all distance values in series to obtain an image Hah; and when similarity is judged, calculating L2 norms of two Hashes, determining that corresponding images are the same if the similarity is less than a set threshold value, otherwise determining that the corresponding images are different. The method is robust for common digital processing such as image rotation, joint photographic experts group (JPEG) compression, noise jamming, brightness regulation and contrast enhancement, and is high in uniqueness.