What are the basic concepts for video data structure and processing

Question

The video file is a lot of images compressed and played at some predefined rate. Of course,  this is a simple and brief description. I need to make a distributed application to process a video in a Hadoop cluster to apply video filters. My question is: Is it necessary to uncompress the video in all images and apply the filter to ALL the images or have a method more intelligent to do this to reduce the time to process it?

Answer 1

1.  Make use of hadoop storm framework. ( It provides to stream the data in real time).

2. The data that is processed through workflow which is called as topology consists of set of sources called spouts and logic of units called bolts.

3. So in this case, you can stream the video through spout and then filter the video using bolts.

4. This unit can encapsulate some video processing logic using OpenCV library for example. 

5.You can use FrameGrabber to parse your video or chunks of video and perform certain filtering steps.