This work presents a patent technique for digital image compression using wavelet
theory and a Huffman CODEC system that can be used for bandwidth reduction in digital
transmission and to decrease time of data transmission in channel information. A twodimensional
1-channel multiresolution analysis and synthesis structure are implemented. In
the transmission, the digital image is processed using quincunx lowpass filtering and quincunx
downsampling, iteratively, in the analysis part, followed by a Huffman encode, in order
to make easy the binary code ordering and to set it for transmission in channel information.
Once binary code is set, it can be sent through a transmission channel. In the reception, the digital
subimage is Huffman decoded and then submitted to the synthesis process, using quincunx upsampling and
quincunx lowpass filtering, multiplying the resulting image by |det Mq|, iteratively, in order to get the digital
image perfect reconstruction. The major contributions of this work are the high rates of compression at the
end of decomposition process, the facility of Huffman codec implementation, comprising the encoding part,
the arrangement of the image vector in the last level, setting it for fast digital transmission through an information
channel and the Huffman decoding part. And finally, the efficiency of the digital images perfect reconstruction.
Simulations using “MATLAB” were developed with digital images in multiresolution, from the
first level until to reach the vector image (last level) in the decomposition part and similarly, from the last level
until to reach the first level, in the image synthesis part (reconstruction). Three different images, with different
gray levels distributions were used as information source. The images are monochromatic, 8-bits and
their dimensions are 1024x1024 (Saturn rings), 512x512 (Lena) and 128x128 (Saturn).