Scribd
Upload
57 views8 pages

EYW4 User Tutorial 2

This document provides an overview of several patches in EyesWeb that analyze physical and geometric parameters extracted from moving images. The patches described calculate the baricenter, bounding rectangle, area, contraction index, and quantity of motion of segmented images. For each parameter, the document explains which EyesWeb blocks are used to extract and visualize the information, and provides an example of possible output videos. It notes that some blocks used are from the legacy EyesWeb 3.2 library and may not be present in the 4.5 distribution at the time of writing.

Uploaded by

Cristian Sandoval Villegas
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
57 views8 pages

EYW4 User Tutorial 2

This document provides an overview of several patches in EyesWeb that analyze physical and geometric parameters extracted from moving images. The patches described calculate the baricenter, bounding rectangle, area, contraction index, and quantity of motion of segmented images. For each parameter, the document explains which EyesWeb blocks are used to extract and visualize the information, and provides an example of possible output videos. It notes that some blocks used are from the legacy EyesWeb 3.2 library and may not be present in the 4.5 distribution at the time of writing.

Uploaded by

Cristian Sandoval Villegas
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 8

DIST - University of Genova Laboratorio di Informatica Musicale (InfoMus) http://www.infomus.dist.unige.

it

EyesWeb Users Tutorial 2


(NOTE: all these patches are based on the Background Subtraction with Multiple Threshold patch analysed in the previous tutorial) (meaning that the Multiple Threshold block used is still the one from the 3.2 distribution) These patches are used to analyse physical and geometric parameters that can be extracted from moving images and that could find an application in fields like medical rehabilitation. The patches analysed are: Baricenter BoundingRectangle Area ContractionIndex QuantityOfMotion

Baricenter

(Up to the median filter the patch is exactly like a Background Subtraction with Multiple Thresholds) The patch shows a possible application of the Baricenter block (Image Operations Baricenter) which will calculate the position of the baricenter of an inputed segmented image. To visualise such information the value has to be processed first through an Ellipse Generator (Geometric Input Ellipse2Dgenerator) and then through a Draw Graphic Object block (Image Operations Draw Graphic Object) which will actually add a representation of our baricenter to our feed. Here is an example of a possible output video:

BoundingRectangle

The patch surrounds our subject with a moving rectangle. To do this we have to use two blocks from the EyesWeb library: the Bounding Rectangle (Image Operations BoundingRect) and the Rectangle Generator block (Input Rectangle2DGenerator). The first will extract the rectangle surrounding the subjects silhouette, while the second one (along with the Draw Graphic Object block that we have already seen) will visualise a rectangle with said coordinates. Here is an example of a possible output video:

Area

In this patch we use the Contraction Index block (Image FeatureCalc ContractionIndex). It returns three outputs: the coordinates of the bounding rectangle (just like the Bounding Rectangle block), the area of said rectangle and its contraction index (i.e. the ratio between the area of the silhouette and the area of the rectangle). In this patch we will be using only the first two outputs, while the third one will be used in the next patch. We visualise our results through a 2-dimensional Cartesian Graph (Math->Scalar->Output>Graph->XvsTime) which plots the area value in time. The bounding rectangle is (Imaging Draw 2PointElement). represented through a 2-Point Element block

Here is an example of a possible output video:

(NOTE: if the processed video presents some artefacts or spurious pixels then the area value will result unprecise) (NOTE2: the Contraction Index, 2-dimensional Cartesian Graph and 2Point Element blocks used are the legacy versions belongng to the EyesWeb 3.2 library and while this tutorial is being written (January 2007) they are not as yet present in the 4.5 distribution)

ContractionIndex

As previously pointed out the contraction index is the ratio between the subjects silhouette area and the rectangles area and represents how the subject utilises the space it is occupying. Thus the index varies between 1 (when the subjects limbs are held close together) and 0 (when the subjects limbs are spread apart). The patch is fundamentally the same as the previous one, so we will only explain how the contration index output is displayed. (which is quite simple, but will be done to maintain coherency with the rest of the tutorial). The output is first processed through an Invalid Handler block (Math Scalar InvalidHandler), which will convert invalid values (e.g. if the subject is not on the screen) to valid ones (in the previous example it would be corrected to 0). The values are then plotted through a 2-dimensional Cartesian Graph, introduced in the previous patch. Here is an example of a possible output video:

(NOTE: the Contraction Index, 2-dimensional Cartesian Graph, Invalid Handler and 2Point Element blocks used are the legacy versions belongng to the EyesWeb 3.2 library and while this tutorial is being written (January 2007) they are not as yet present in the 4.5 distribution)

QuantityOfMotion

The patch shows how the Motion History Image block (Image Effects MHI) can be used. The block returns two outputs: the actual SMI (i.e. a silhouette image that carries with it information about any variations in shape or movement in the inputted silhouette) and the Quantity of Motion (i.e. the area of said SMI). Both these informations can be used to measure and detect overall amounts of motion in a video feed. Visualisation is quite simple and straight-forward: the SMI added to the original input feed and displayed by usual means, while the Quantity of Motion is plotted through a 2-dimensiona Cartesian Graph. Here is an example of a possible output video:

(NOTE: the 2-dimensional Cartesian Graph block used is the legacy versions belonging to the EyesWeb 3.2 library and while this tutorial is being written (January 2007) it is not as yet present in the 4.5 distribution) References: Antonio Camurri, Barbara Mazzarino, Gualtiero Volpe, A tool for analysis of expressive gestures: The EyesWeb Expressive Gesture Processing Library.

You might also like