University of California

Samsung Electronics Co., Ltd.

Sanjit K Mitra

Dong-seek Park

Bangalore S Manjunath

Rossano Marchesani

Kye-ho Park

Jong-dae Kim

Sung-Hwan Jung

Eduardo Abreu

Debargha Mukherjee

Jong Jin Chae

A method for digital watermarking and, in particular, for digital data hiding of significant amounts of data in images and video. The method employs a discrete wavelet transform for embedding gray scale images which can be as great as 25% of the host image data. A simple control parameter is used that can be tailored to either hiding or watermarking purposes, and is robust to operations such as JPEG compression. The method also uses noise-resilient channel codes based on multidimensional lattices which can provide for embedding signature data such as gray-scale or color images. Furthermore, embedded image data can be recovered in the absence of the original host image by inserting the data into the host image in the DCT domain by encoding the signature DCT coefficients using a lattice coding scheme before embedding, checking each block of host DCT coefficients for its texture content, and appropriately inserting the signatured codes depending on a local texture measure. The method further provides for source coding the signature data by vector quantization, where the indices are embedded in the host by perturbing it using orthogonal transform domain vector perturbations. The transform coefficients of the parent data are grouped into vectors, and the vectors are perturbed using noise-resilient channel codes derived from multidimensional lattices. The perturbations are constrained by a maximum allowable mean-squared error that can be introduced in the host. Also, speech can be hidden in video by wavelet transforming the host video frame by frame, and perturbing vectors of coefficients using lattice channel codes to represent hidden vector quantized speech. The embedded video is subjected to H.263 compression before retrieving the hidden speech.

Method for embedding and extracting digital data in images and video

A local relaxation method for estimating optical flow using a Poisson equation is provided. The local relaxation method includes the steps of a) determining the Poisson equation in a linear matrix of Gx=Ex+b according to a change of time, b) decoupling the determined velocity combination term of the Poisson equation, and c) applying a displaced frame difference of a temporal change to the Poisson equation where the velocity combination term is decoupled. Accordingly, the convergence speed is faster than that obtained using conventional technology. Also, in the case that a displaced frame difference is applied, the convergence value is accurate and motion in every pixel can be estimated in order to enhance image quality, and in the case of a video code, compression gain can be obtained by using a portion of a motion vector, i.e., the motion vector of a low band signal.

Local relaxation method for estimating optical flow

A method and apparatus for shaping the impulse response of a non-minimum phase communications channel, by reversing the CIR without enhancing the noise level using non-causal allpass filters is disclosed. Unstable allpass filters are implemented as stable non-causal filters operating in reversed time to obtain a minimum phase response from a non-minimum phase CIR. The original CIR is estimated using adaptive algorithms, and the coefficients of the allpass filters are taken directly from the estimated CIR. The cascade of the channel and allpass filter has an impulse response that is the approximate time reversed version of the original CIR. A 'block' allpass equalizer and an 'optimal' allpass equalizer are used in a two-stage filter to increase system performance. Other multiple stage filter configurations are disclosed, which can include use of decision feedback equalizer (DFE) decoders as well as Viterbi decoders. The invention is effective in the case of two-path channels, but can also give satisfactory results when more paths are present. Moreover, the present invention is well suited to 'sharpening' the CIR with only minor variations in noise level. A minimum phase desired impulse response (DIR) can be obtained with only a few nonzero samples, leading to a substantial reduction in complexity of the Viterbi algorithm, with only minor variations in the noise level.

Method and apparatus for multipath channel shaping

A fast discrete cosine transform (DCT) based method of encoding and decoding signals is disclosed. A signal of finite length is decomposed into subsignals of shorter length and an approximate DCT (A-DCT) is computed by keeping only selected subsignals. Computation time decreases relative to the number of subsignals being used to compute the A-DCT, fewer subsignals resulting in a faster computation. For signal compression applications, the computed A-DCT coefficients are quantized and coded. For signal recovery, the encoded A-DCT coefficients are processed through a decoder and the decoded signal is recovered using a direct inverse DCT (I-DCT). In some applications, an approximate inverse DCT (AI-DCT) based on similar decomposition can be used to recover the signal from the encoded data. In image coding applications, the image recovered from the decoded signal exhibits negligible block artifacts compared to conventional JPEG, particularly for low bit rate compression.

Discrete cosine transform-based image coding and decoding method

Growing community of inventors

Santa Barbara, CA, United States of America

Sanjit K Mitra