In this work, we constructed various algorithms in order to extract frames from videos using OpenCV Computer Vision library. The results are saved as a text file as well as a video file that consists of the extracted frames only.

Here you can download the text for the thesis. (It is written in the Greek language only)

Requirements: Opencv library

You can use the Makefile in order to compile all the algorithms at once

Usage: make all                 # compile all algorithms
       make <algorithm_name>    # compile the source code
                                # for a specific algorithm
       make clean               # delete any object files
       make help                # Print this help message

where algorithm_name choose one of the following:

• absdiff
• comprdiff
• ecrdiff_v1
• ecrdiff_v2
• histdiff

This Makefile calls a series of other Makefiles in order to compile all the source code

For each algorithm there is a set of executables

• <algorithm_name>_export_images: Exports video frames as images in PNG format (marked only). This images can be pass to ffmpeg in order to construct the resulting video.
  • ./<algorithm_name>_export_images <input_video_file> <output_TXT_file>
• <algorithm_name>_pipe_pixels: Sends BGR (24 bpp) frames to stdout. For example, you can pipe the output to ffmpeg in order to write video having more choices and control over the encoding procedure.
  • ./<algorithm_name>_pipe_pixels <input_video_file> <output_TXT_file>
• <algorithm_name>_write_video: Write marked frames in a video files using OpenCV. The resulting video includes only marked frames and has the duration of the original video.
  • ./<algorithm_name>_write_video <input_video_file> <output_video_file> <output_TXT_file>

The <output_TXT_file> contains the results of the execution of the algorithm, i.e. marked frames, video stats etc.

##Platforms

• Linux
• Windows

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.