Considering the character of lane surveillance
system, this paper proposes a special algorithm for the
application of lane surveillance system. This proposed
algorithm processes the shot segmentation and the key
frame extraction before video compression. What is
more, it provides correspond information for the latter
video compression. In detail, it provides the extracted
key frame as the I frame for video compression, the
one precedes the key frame as the B frame, and the
one follows the key frame as the P frame. In this way,
we can obtain the most useful amount of information
while keeping the storage to hold the video
compression low. When we want to find out the useful
information from the huge volume of video data, we
can greatly save the time spent on video retrieval,
because we directly use the I frame as the key frame to
retrieve. Subsequently, we can get the clearest
decompression image when we decompress the video
data.