We propose an image hashing algorithm that is based on distributed compression principles. The algorithm assumes the availability of a robust feature vector extracted from the image. Then a suitable dither sequence is added to this feature vector, and the resulting dithered feature vector is compressed using distributed compression principles. We prove that the dither sequence can be used to guarantee information-theoretic security. Applications of our proposed secure image hashing scheme include video watermarking, image authentication, and image database management.