초록 ▼

To permit scaling or rotation of an image to be discerned--even if the original image is not available--certain marker signals are steganographically encoded in the image. In one particular implementation, encoding is effected by adding an overlay signal to the image. The overlay signal--when repres

To permit scaling or rotation of an image to be discerned--even if the original image is not available--certain marker signals are steganographically encoded in the image. In one particular implementation, encoding is effected by adding an overlay signal to the image. The overlay signal--when represented in the spatial frequency domain--includes a pattern of spots. If the image is scaled or rotated, this pattern changes in known ways. By examining the change in this pattern, the distortion (i.e., scale/rotation) of the image can be inferred. This technology finds particular application in conjunction with watermarking systems, permitting recovery of watermark data from images that have been scaled or rotated subsequent to watermarking.

대표청구항 ▼

1. A method of processing an image that has been subject to at least one of scaling or rotation from an initial state, the method comprising: discerning a known pattern of frequency components from a steganographic watermark in said image; and by reference to a size or rotation of said pattern f

1. A method of processing an image that has been subject to at least one of scaling or rotation from an initial state, the method comprising: discerning a known pattern of frequency components from a steganographic watermark in said image; and by reference to a size or rotation of said pattern from an expected configuration, determining scaling or rotation of said image from an original state. 2. A method of steganographically decoding an encoded image to extract plural bits of auxiliary data secretly conveyed therein, characterized in that the auxiliary data can be recovered from the encoded image without reference to an unencoded image, and can be recovered even after rotation and scaling of the encoded image, the method including finding a pattern in a transform domain that corresponds to a characteristic pattern, and discerning a transformation of said found pattern from said characteristic pattern. 3. The method of claim 2 in which said transform domain is the spatial frequency domain. 4. The method of claim 2 in which the method includes use of the Fourier-Mellin transform. 5. The method of claim 2 in which said patterns comprise patterns of energy peaks in the transform domain. 6. The method of claim 2 that further includes, from said transformation, determining rotation and scaling of the encoded image.