SIGVR Description.

  Before describing the opportunities shared by SIGVR, let’s take a brief look to some well known ways for condensation of texts. This will lead us to a better understanding of the tasks performed by SIGVR when it processes input video sequences.

   The information to be presented is also available in the following promotional presentation:

   Every document that develops some particular topic should also contain a summation in which is presented to the reader, in a very concise and brief way, its purpose. This summation, by convention, should have length by one paragraph and it is precisely called Abstract.

   On the other hand, a Summary is seen as a text derived from another manuscript. A Summary should present in explicit way the main ideas exposed in the source text. Regularly a Summary’s length should not exceed by a third part of the original text.

   The abovementioned implies, in terms of extension, that an Abstract is smaller than a Summary.

   Following this line of reasoning, it is well known many readers extract the main idea from every paragraph in a text. This idea is materialized in a main sentence which is directly associated to its corresponding paragraph. Therefore, this set of sentences could describe the basic and essential structure of a whole text.

   Now well, we know a DNA molecule (DeoxyriboNucleic Acid) contains basic and fundamental descriptions about the structure and functioning of a living organism. Analogously, we establish the set of all main sentences extracted from each paragraph gives form to the DNA Text Sequence for a text being condensed.

   In this sense, we have three ways for text condensation sorted according to their length:

• Abstract.
• Summary.
• DNA Text Sequence.

   All right but, what is SIGVR? What is done by SIGVR? The Intelligent Software for Video Summaries Generation (SIGVR) is the result of a combination of state-of-the-art techniques that are being part of the following areas:

• Computational Intelligence.
• Computer Vision.
• Signal Processing.
• Computational Geometry.

   For a better understanding it is useful to take in account the following points:

• A video sequence captures a set of events.
• Depending on the video sequence’s length and the context, there could be captured a great amount of events.
• Some events could be characterized as important, significant, or substantial, while others could be omitted without affecting the purpose or information presented in the video sequence.

   SIGVR is a patenting-in-process computer system that analyses a video sequence and identifies its significant and substantial events. These events are extracted and condensed in a muted-audio new video sequence which corresponds to a summarized version of the original video sequence.

   SIGVR provides, for a video sequence, three ways of condensation sorted according to the number of identified events:

• The Video Abstract.
• The Video Summary.
• The DNA Video Sequence.

   Such as happens with written texts, the Video Abstract represents an instance of condensation where there are identified a minimal quantity of significant events but they are enough for understanding the content of the video sequence.

   In the Video Summary’s case, SIGVR identifies and condensates, for a given video sequence, all events that were characterized as significant.

   At last, for an input video sequence, SIGVR is capable of making up its DNA Video Sequence by considering all relevant and non-relevant events.

   Unlike Video Abstract and Video Summary, SIGVR gives form to the DNA Video Sequence by taking only one frame for each identified event. This frame is called Representative Frame.

   From the User’s point of view, SIGVR only receives as input the video sequence to be processed and provides, as output, according to the selected condensation mode, a new video sequence (in the cases for Video Abstract and Video Summary) or a set of representative frames (in the case of DNA Video Sequence).