| Summary: | 碩士 === 國立中正大學 === 資訊工程研究所 === 82 === Pyramid images are an important class of progressive image
transmission techniques, in which different levels correspond
to a series of reduced-resolution approximations of the
original image. In this study, the detection and correction
approaches to transmission errors in pyramid images using two
kinds of variable-length codes are proposed. For pyramid images
using variable-length codes, a transmission error in a codeword
may cause the subsequent codewords to be misinterpreted,
resulting in loss of synchronization of the decoder. Here two
variable-length codes with self-synchronizing codewords and the
universal synchronizing sequence (the EOL codeword),
respectively, are used to encode pyramid images so that the
synchronization problem can be solved. Both the two types of
pyramid images treated in this study, namely, the difference
pyramid and the modified reduced-difference pyramid, have the
level-independent property, i.e., transmission errors induced
at higher levels will not affect the error detection and
correction operations performed at lower levels. The objective
of the proposed approaches is to recover high-quality pyramid
images from their corresponding corrupted pyramid images. The
proposed error detection and correction approaches are based on
two corresponding checking conditions for a block containing
four nodes in the two types of pyramid. When four codewords
within a block violating the corresponding checking condition
are detected, a sequence of single-bit-inversions and
redecoding processes are performed on the block and/or the
previous blocks until a feasible solution is found. Based on
experimental results, the proposed approaches can almost
recover the original or high-quality pyramid images from their
corresponding corrupted pyramid images. This shows the
feasibility of the proposed approaches.
|
|---|
