Efficient detection of error blocks in a DCT-based compressed video sequence

摘要

A method of efficient and reliable detection of error blocks in a DCT-based compressed video sequence extracts edges from an image that correspond to the DCT blocks. These edges are processed to determine an edge energy value. The edge energy value is compared with a threshold to provide an alarm for an error block when the threshold is exceeded. The edge energies for each block may be summed and compared with an overall threshold value to generate the alarm, or the edge energy for each edge may be compared with an edge threshold value to determine which are good and which are suspect, with the alarm being set when at least three of the edges are suspect (two edges for corner blocks).