.TH QCCTCEENCODE 3 "QCCPACK" "".SH NAMEQccWAVtceEncode, QccWAVtceDecode \-encode/decode an image using the TCE algorithm.SH SYNOPSIS.B #include "libQccPack.h".sp.BI "int QccWAVtceEncode(const QccIMGImageComponent *" image ", int " num_levels ", int " target_bit_cnt ", double " stepsize ", const QccWAVWavelet *" wavelet ", QccBitBuffer *" output_buffer );.br.BI "int QccWAVtceDecodeHeader(QccBitBuffer *" input_buffer ", int *" num_levels ", int *" num_rows ", int *" num_cols ", double *" image_mean ", double *" stepsize ", int *" max_coefficient_bits );.br.BI "int QccWAVtceDecode(QccBitBuffer *" input_buffer ", QccIMGImageComponent *" image ", int " num_levels ", const QccWAVWavelet *" wavelet ", double " image_mean ", double " stepsize ", int " max_coefficient_bits ", int " target_bit_cnt );.SH DESCRIPTION.SS Encoding.LP.B QccWAVtceEncode()encodes an image component,.IR image ,using the TCE algorithm by Tian and Hemami.The TCE (tarp coding using classification to achieve embedding) algorithmis based on the tarp algorithm (see .BR QccWAVTarpEncode (3)),but is designed to provide better rate-distortion performancewhen used in an embedded fashion; i.e., when decoding is performedat a rate less than that at which the bitstream was produced.See "ALGORITHM" below for more detail..LP.I imageis the image component to be coded and.I output_bufferis the output bitstream..I output_buffermust be of.B QCCBITBUFFER_OUTPUTtype and opened via a prior call to.BR QccBitBufferStart (3)..LP.I num_levelsgives the number of levels of dyadic wavelet decomposition to perform,and.I waveletis the wavelet to use for decomposition..LPTwo encoding modes are supported.Mode 1 consists of first applying a scalar quantizer to allDWT coefficients and then encoding the coefficients,bitplane by bitplane, until all bitplanes are encoded.Mode 2 consists of direct bitplane encoding without anyscalar quantization of coefficients; in this mode, encodingstops when the target bitrate is reached.Mode 1 is selected by specifying a quantization.I stepsizethat is greater than zero.If.I stepsizeis less than or equal to zero, thenMode 2 is selected. In this case, a target encoding bit count,.I target_bit_cntmust be specified..SS Decoding.LP.B QccWAVtceDecodeHeader()decodes the header information in the bitstream in the input.I input_buffer(which must be of.B QCCBITBUFFER_INPUTtype and opened via a prior call to.BR QccBitBufferStart (3)).The header information is returned in.I num_levels(number of levels of wavelet decomposition),.I num_rows(vertical size of image),.I num_cols(horizontal size of image),.I image_mean(the mean value of the original image),.I stepsize(quantization stepsize), and.I max_coefficient_bits(the number of bits used to represent the magnitude of the coefficient with largest absolute value)..LP.B QccWAVtceDecode()decodes the bitstream.IR input_buffer ,producing the reconstructed image component,.IR image .The bitstream must already have had its header read by a prior callto.B QccWAVtceDecodeHeader()(i.e., you call.B QccWAVtceDecodeHeader() first and then.BR QccWAVtceDecode() )..SH "ALGORITHM"It is widely believed that, in the wavelet domain, coefficients in a local neighborhoodcapture essential context information, such as edges and patterns, and thatthis information can facilitate compression.Many image coders exploit such local information in the form of adaptive entropy coding.Alternatively, tarp filtering (see.BR QccWAVtarpEncode (3))produces probability estimates through an IIR filtering techniqueon the bitplanes of  the wavelet coefficients. While providing very good performance when used in a nonembedded manner, the original tarp coder performs less competitively when used in an embedded manner, in spite of its operation on bitplanes.The reason is that the underlying tarp filter is designed to followa raster-scan encoding order;however, the use of fractional bitplanes (see Ordentlich.IR "et al" .)provides better rate-distortion performance for embedding. Such fractionalbitplanes require an encoding order more flexible than a simple raster scan.  In the TCE algorithm,coefficients are classified according to statistical properties, andthe tarp filter is run on only the single class on which it tends to generate accurate probability estimates..LPIn the TCE algorithm,.I nonzero-parentcoefficients and.I runcoefficients from the fractional-bitplane approach of Ordentlich.IR "et al" .are combined to form a single class, the.I zero-runcoefficients. For each bitplane, the TCE system then performsthree passes:  1) adaptive arithmetic coding of bits of.I nonzero-neighborcoefficients; 2) tarp filtering and non-adaptive arithmetic coding of.I zero-runcoefficients; and 3) encoding of refinement bits with an adaptive arithmetic coder.The sign bits of coefficients are coded when needed with a probability of 0.5. The encoding/decoding ends when the target rate is reached. Tian and Hemami describe several approaches to improving the probabilityestimate of the tarp filtering of the second pass..SH "SEE ALSO".BR tceencode (1),.BR tcedecode (1),.BR imgdwt (1),.BR imgdist (1),.BR QccBitBuffer (3),.BR QccENTArithmeticEncode (3),.BR QccENTArithmeticDecode (3),.BR QccWAVSubbandPyramid (3),.BR QccPackWAV (3),.BR QccPackIMG (3),.BR QccPack (3).LPC. Tian and S. S. Hemami, "An Embedded Image Coding SystemBased on Tarp Filter with Classification," in.IR "Proceedings of the International Conference on Acoustics, Speech, and Signal Processing" ,Montreal, Quebec, Canada, May 2004, vol. 3, pp. 49-52.E. Ordentlich, M. Weinberger, and G. Seroussi,"A Low-Complexity Modeling Approach for Embedded Coding ofWavelet Coefficients," in.IR "Proceedings of the IEEE Data Compression Conference" ,Snowbird, UT, March 2002, pp. 23-32..SH AUTHORWritten by Chao Tian, Cornell UniversityCopyright (C) 1997-2005  James E. Fowler.\"  The programs herein are free software; you can redistribute them and/or.\"  modify them under the terms of the GNU General Public License.\"  as published by the Free Software Foundation; either version 2.\"  of the License, or (at your option) any later version..\"  .\"  These programs are distributed in the hope that they will be useful,.\"  but WITHOUT ANY WARRANTY; without even the implied warranty of.\"  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the.\"  GNU General Public License for more details..\"  .\"  You should have received a copy of the GNU General Public License.\"  along with these programs; if not, write to the Free Software.\"  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.