A method for coding an image or other two-dimensional data array to provide a sequence of images having differing spatial frequency content is disclosed. The method generates the output images by taking weighted sums of the pixels in the input image. The weights are sets of two-dimensional irreducible scaling and wavelet coefficients.
A method and apparatus for adaptive bit allocation and hybrid lossless entropy encoding in a lossy compression system. The invention includes three components: (1) a transform stage to decorrelate image data into a baseband and multiple subbands, (2) a quantization stage to quantize the resulting transform coefficients, and (3) a lossless entropy coder stage to encode the quantized indexes. In the preferred embodiment, the transform stage uses a wavelet transform algorithm. The quantization stage adaptively estimates values for parameters defining an approximation between quantization size and the logarithm of quantization error, and recursively calculates the optimal quantization size for each band to achieve a desired bit rate. The baseband and subbands are transformed into quantization matrices using the corresponding quantization sizes. The lossless entropy coder stage uses the observation that the entropy property of run lengths of zero index values in the subband quantization matrices is different from the entropy property of non-zero indices. Each quantization matrix is parsed so that each non-zero index is extracted into a separate stream, and the remaining position information is parsed into an odd stream of run length values for "0" and an even stream of run length values for "1". These three streams are Huffman coded separately in conventional fashion. This hybrid algorithm gives an approximately 10% percent improvement over conventional run length and Huffman coding for similar images. The overall compression algorithm gives about 2.about.6 dB improvement in terms of peak signal-to-noise ratio (PSNR) over JPEG algorithms for similar images at similar bit rates.
The invention relates to an image sequence coding method in which images are segmented and coded with respect to their contours and textures. The texture coding step is carried out by means of a new technique relying on a wavelet decomposition of the images, called quincunx bidimensional wavelet transform and adapted to a region-based coding scheme, for applications allowing to reach very low bit rates while keeping a good image quality.
An image processing apparatus for orthogonally transforming image information in a block comprising NXN pixels, and creating block information composed of MXM pixels (N.noteq.M) on the basis of the information orthogonally transformed. The created block information is reverse transformed into real spatial information having MXM pixels. The reverse transformation is performed using a basic vector other than a transposed matrix of an orthogonal transformation matrix of MXM pixels.
A system and method implementing a fast wavelet transform by shifting a pair of pixels into a single shift register and using a multimode quadrature mirror filter to eliminate the need of downsampling the filtered signals and to decrease the area required to implement the device on a semiconductor chip.
An analog circuit implementing a continuous wavelet transform forms a multiplicity of analog wavelet outputs. These analog wavelet outputs can be used for data compression. This analog circuit is estimated to be about one-hundredth (1/100) the size and power of a digital wavelet transform circuit. Additionally, the analog wavelet outputs of the analog wavelet transform chip is directly determined without the loss of information due to the digital sampling.
