* Using environment variable to store options parsed by av_dict_parse_string(ENV{OPENCV_FFMPEG_CAPTURE_OPTIONS}, ";", "|")

* Adding missing mandatory flags parameter

* Guarding against missing function via LIBAVUTIL version

* Code review fixes

Copy/paste error due to coder mistake reverted
Proper version checking for LIBAVUTIL_BUILD

Prevents 10000x1 => 10000x8 transformation after getContinuousSize() call

Incorrect Naming (#9599)

* Resolved #9544

* This is a better fix

* This should be good.

Input arrays must be depth CV_8U and of identical size.

fixed unnecessary reconstruction in seamlessclone by using only the
required rectangular area in input images