The Project is about floor care, to be more precise about processes in vacuum cleaning. It is based on research Philips made 16 years ago, which gave the initial idea for this new Project. In the former project, Philips evaluated the effects of direct application of forces to the carpet. In this case the whole carpet was accelerated with a vibration exciter, instead of accelerating the air by the use of a vacuum cleaner to cause forces on the sand particles.
The aim of this new project is to gain more knowledge about the relations between dirt and carpet. The main questions of the project are:
What are the causes that keep the dirt in the carpet?
How much energy is needed to remove the dirt from the carpet?
How big are the forces applied with a common vacuum-cleaner
The reasons why sand stuck in the carpet are for example caging effects in which the dirty is caged by carpet fibers or bound to carpet fibers by water adhesion.
Therefore an experimental setup is needed to be developed that could evaluate the effects that bind the dirt to the carpet.
This led to the question what are the suitable applications to evaluate these issues.
The decision was made to use of a high-speed camera in combination with a particle image tracking software package to analyze the trajectories of the sand particles.
In the previous experiments it was measured how much sand does leave the carpet, while applying a certain force, like rotating a piece of carpet at certain velocity or moving it with a vibration exciter at a certain frequency.
The new experimental setup is designed to capture images, with the high-speed camera, from the sand while leaving carpet due to applied forces of a vibration exciter. The image recordings are made from different directions.
The captured images are analyzed by the particle image tracking software package to track the particles while leaving the carpet. The software package will also give information about the approximate size of the sand particles and their acceleration and velocity.
Also, a Matlab program was written for more detailed analysis of the results of the particle image tracking software. The Matlab code does for instance sort out particles with inadequate amount of images in which they were tracked, as well as particles that do appear bigger or smaller in the images as the determined size given by the safety chart of the used material.
The following chapters will give detailed information about the used hardware and software equipment.
The experimental setup is designed to measure how many percent of sand has left the carpet after applying a constant alternating force for duration of 30 seconds, and to measure the velocity of the sand while leaving the carpet for being able to conclude the force acting on the sand particle. The results will be statistically analyzed by the software for behavior estimation of the sand particle.
Camera and Camera software
The High-speed Camera
The main tool for the experiments was a RedLake IDT Motion Pro X4 high speed camera. This camera has the capability to record movies of up to 5130 frames per second with a resolution of 512x512 pixels. The size of a single pixel on the camera chip is 16 micrometers and the distance between the centers of the pixel is also 16 micrometer. The minimal exposure time is 1 microsecond. The frame rate determines the exposed time an image has, so this means the frame rate can be used as a time scale. Attached to the Camera was a Leica Z6 APO lens with a 0.8x C-Mount. The Z6 APO lens has variable magnification of 0.57x to 3.6x. When mounted on the camera the lens has a minimum length of 168.69 mm. This lens is designed to correct the light rays over its complete focal area, so all light rays that the objective collects will be focused in one point. Through this correction parallax errors will be avoided and the record images are evenly sharp from their center to their edges.
Camera software
Redlake IDT provides also the software to control the high speed camera and save its recordings, it is called Motion Pro 4.0.100. All adjustable parameters of the camera are controlled with this software package, like frame rate, exposure time, recording duration and calibration of the image sensor. After starting up the camera, it needs to be calibrated. In the toolbar, the selection Camera are the advanced settings situated. Under this point the camera calibration is listed (Fig. 3-1).
F:\MotionStudio1-2.bmp
Figure 3: Camera calibration option in the Camera software
The camera calibration will subtract the image background to reduce the noise level in the recorded images. The software takes an image without illumination and measures the signal information in this image. The measurement is done to remove this signal information from all images that will be recorded in the upcoming camera session, to reduce the noise level in the images (Fig. 3-2).
F:\MotionStudio2-2.bmp
Figure 3: Settings in the Camera background calibration option
When the camera calibration is done, the image acquisition parameters can be set. On the right side of the camera software window are options to set parameters for the image acquisition. There are two overall choices, "Live" card and "Playback" card.
The in Live card all parameter will be set before the recording starts. The "Live" card offers the opportunity to have a live view through the camera by pressing the play button.
While having a look at the live view it is possible to set the camera frame rate and the exposure time in the "Live" card. To change sensor gain it is necessary to stop the live view and switch to the Playback mode.
F:\MotionStudio3-2.bmp
Figure 3: Live Card in the camera software with live view option, frame rate setting, exposure time setting and sensor gain
F:\MotionStudio4-2.bmp
Figure 3: Warning message after pushing the Rec button
The live view is used to adjust camera in front of experiment object, to place it in focus and optimize the illumination, which will be made just by sight. When the camera is in its optimal position, the experiment can be recorded. Before the second recording the software gives a warning that all previously acquired images will be erased for the following recording (Fig 3-3, 3-4).
After the recording is made, the images need to be transferred from the camera to the computer hard drive. In the toolbar the option file lists the "Save Acquisitions…" option.
F:\MotionStudio5-2.bmp
Figure 3: Save Acquisitions… option in the Software
F:\MotionStudio6-2.bmp
Figure 3: Option to save the images with in the Save Acquisitions option
In this option it will be defined in which folder main folder the image folder will be stored, how the images will be named, as what data type the images will be stored and a "from frame to frame" selection can be made (Fig. 3-5).
The software offers several image types such as .bmp or .tiff which the images will be stored. It also allows combining all images to an avi-movie (Fig. 3-6).
More detailed information about software and camera will be found in the software manual.
Measuring the Size of the particles
In the experiments, moving sand particles were recorded with the high speed camera. A sand particle does occupy a certain amount of pixel in an image of the camera. With the information how many pixel a sand particle does occupy, it is possible to calculate its approximate size and weight. Therefore it is necessary to know how many millimeters or micrometer a pixel does represent. The camera lens does magnify the image, so it is not a one by one image. This means a pixel does not represent its own size on the camera chip it does represent a magnification of the observed area.
The size of a pixel was measured with the Orange-Pixel-Meter, a pixel measurement tool developed by the company "inspiredORANGE" which seems to be expired. But the program still can be downloaded on www.softonic.com. This software basically just measures the amount of pixel between two points in an image. In the first place a caliper was placed in front of the camera lens. It had to be ensured that the caliper is as parallel as possible to the axis of the image. This is needed to avoid parallax errors. Adjusting the ruler is made by hand, and is at its optimum when the ruler appears to sharp at all points in the image.
Also it is needed to ensure that the caliper section which is observed by the camera is in ideal focus. When the caliper is placed correctly in front of the camera, a picture will be taken with the high speed camera.
E:\Experiments\PixelToMillimeterDefinition_001\Totalimagewidth.bmp
Figure 4: Measuring the pixel length of the camera images
E:\Experiments\PixelToMillimeterDefinition_001\measuringlengthofamillimeter.bmp
Figure 4: Measuring the pixel length of a millimeter
The camera takes pictures with a resolution of 512x512 pixels. When the caliper is adjusted to its optimal position beneath the lens, a picture of the caliper will be taken. Then the picture will be measured with the Orange-Pixel-Meter software package. First it is needed to ensure that images have the right size. Therefore the amount of pixels in horizontal and vertical direction will be measured. If the amount of pixels is 512 in the horizontal and vertical direction, the amount of pixel of a millimeter will be measured from the image of the ruler with the measurement tool (Fig 4-1, 4-2).
Hence the pixel on the cameras image chip have the size of a few micrometer, the millimeter scale will be converted to micrometers, to calculate the size of a pixel in the image, so 1000 micrometers will be divided by the amount of pixels for the distance of a millimeter. The result of this will be translated back in millimeter.
Particle Tracking software
This chapter discusses the particle tracking software, which analyses the image set made with the high speed camera. It will be explained how the software will be applied, and which results it will give. Also it will be illustrated what changes were made to the software in order to get more information out the recordings.
The software is developed by Philips, it is programmed in LabView. The software is executed in LabView, so it is not stand-alone software. The major issue with this software is that it has no documentation, or the naming of the parameters are not selfdiscriptive, so it is not possibile to explain in detail how the software does work and not all software settings can be explained or properbly applied. But still it was possible to add some analysis functions to software by Adrie Raijmaaker who is one of the developers of this software.
Detailed Software Guide
F:\ParticleSoftware1.bmp
Figure 5: The main window of particle tracking software
The first window when the program is called in LabView. This view shows four small buttons in the upper left area, directly below the toolbar. Below the four small buttons, five bigger buttons are situated: convert avi ->bmp, generated background, prepare analyse, analyse and save result. Below the five buttons the millimeter/pixel ration and the frame rate can be set. Right from the five bottons there are three big cards: analyse, Background and result. Below the five buttons and the three cards, the folder that contains the images and the folder in which the analysis shall be stored. The three cards will be explained later in this text (Fig. 5-1).
Figure 5: Control buttons in the particle tracking software
The software starts operating, by pushing the arrow button in the upper left corner below the toolbar. To analyze the images, the folder with the images should be set. The first step in analyzing the images is to generate a background image that will be used to denoise the images, similar to the denoising process in the camera software. Pushing the "generated background" button opens the folder with the images, than the first image needs to be selected; now a warning message appears and says a folder with the same name already exists. Push Ok and choose Current folder, this will close the folder and create the background image. Now the analysis can be prepared. First step in this process is to click on the "prepare analyse" button, this opens the folder with the images again. Now select the first image of the series and click Ok, then select the image with the name "background" and click Ok again and the folder closes. Then frame rate of the image series and the pixel to millimeter ratio needs be set, to get the right result of the image analysis. These settings are below the five buttons on the left side of the main window (Fig. 5-3).
F:\ParticleSoftware-Screenshots\ParticleAnalysis7.bmp
Figure 5: Warning message while creating the background image
The next step is to adjust the parameters for the analysis. The analysis will be adjusted in the analyse card. On the left side of the analyse card are the settings that need to be adjusted for the analysis, on the right side are two images, the original image on top and the processed image below.
Figure 5: Software settings to optimize the particle tracking analysis
On the left side are the color settings: these define if the images are grayscale or RGB color. Below the color settings are the idiff settings, to define which minimal and maximal size the particle should have that will be tracked, with "ldiff detection area (pixels)" the minimal amount of pixels will be set, and with upper the maximal amount. These settings will avoid that the tracking program will track particle that got out of focus and hence got an irrational size. On the right side from the ldiff detection area the settings for erode and dilate iterations,that calculate a more realistic shape of the particles. These calculations are necessary because of the camera sensors: as the camera sensors are square shaped, the particles appear with serrated edges. Beneath the detection area and iteration settings, the threshold can be set (Fig. 5-4).
During adjusting the settings the analyze card shows in the "result image" changes in its appearance, so with ideal settings one will see red dots in on the same position in the picture as were particles are in the original image above and nowhere else.
F:\ParticleSoftware-Screenshots\ParticleAnalysis3.bmp
Figure 5: Unprocessed image on top with tracking markers and processed image in the bottom
F:\ParticleSoftware-Screenshots\ParticleAnalysis5.bmp
Figure 5: Background image on the left and actual image in the analysis process on the right
After the settings are adjusted the program can analyze the images. Pushing the analysis button does start the analysis. During the analysis it is possible to observe the difference between the original image and the processed image in the analyze card, between the background image and the actual image (Fig. 5-5, 5-6).
F:\ParticleSoftware-Screenshots\ParticleAnalysis6.bmp
Figure 5: Actual image on upper left side and trajectory graphs in a XY axis
The actual image and an image of a XY coordinate that draws the tracks of the particles in the result card. The background card has just observing purposes. The tracks of the particles in XY Graph are drawn from the same origin. This means that the XY graph does display the trajectory of the particle but not its point of origin within the recorded images. In the result card it is possible to set that just particle from a certain chain of tracks should be displayed in the XY coordinates. This option is below the XY graph. Below the actual image is a list with all tracks the software recorded, this list is not affected by the settings for the minimum chain. This list consists of six columns: track, start frame, end frame, chain, distance (pixels). After the recording it is possible to select a particular track by clicking on its track number in the list and this track will be displayed in the XY coordinates and the last image in which it appeared will be displayed. At the end of the analysis the image of the XY coordinates can be saved (Fig. 5-7).
Image Tracking Results
After the analysis the gathered information can be saved be clicking on the "safe result" button. This creates a text file that stores all gathered information as well as all parameters applied during the analysis (Table 1). In text file are two tables, the table called "tracks" which contains Meta information about the particles, and the table "subtracks" which contains detailed information about the particles.
[General]
File Name=E:\\Temp\\beschleunigung\\01.07.2010.txt
Program.Name=
Program.Version=0.00
Created="01-07-2010 14:34:13"
Operator=
SEP Version=2.00
[Comment]
[settings]
UI.ldiff detection area (pixels).Range Lower Value=65.000000
UI.ldiff detection area (pixels).Range Upper Value=10000.000000
UI.dilate iterations=2
UI.erode iterations=2
UI.color=Grayscale
UI.threshold.method (false)=TRUE
UI.threshold.level.Lower=9.990000
UI.threshold.level.Upper=111.000000
UI.threshold.contrast enhancement factor=1.000000
UI.threshold.area (%)=98.800003
ROI Descriptor.Global Rectangle="<size(s)=0> "
ROI Descriptor.Contours.<size(s)>=0
[tracks]
;track start frame subtrack start subtrack stop total distance (pixels) mean velocity (mm/sec)
1.00 1.00 1.00 3.00 0.54 361.02
2.00 1.00 4.00 122.00 62.61 1052.20
3.00 1.00 123.00 125.00 0.66 440.95
. . . . . .
. . . . . .
[subtracks]
subtrack track frame X Y distance velocity acceleration area
1.00 1.00 1.00 76.86 -49.12 -1.00 0.00 0.00 72.00
2.00 1.00 2.00 76.57 -49.00 0.32 635.04 1270081.16 65.00
3.00 1.00 3.00 76.51 -49.22 0.22 448.01 -374067.63 65.00
. . . . . . . . .
. . . . . . . . .
Table 1: Output file with information about software settings and collected information
Throughout the whole analysis, it is possible to illustrate where the particles in the image are which software tracks by choosing the "Display marker" option. This also indicates if the settings of the tracking software are applied properly. For instance if there is clearly visible particle which is not tracked, the software settings need to be readjusted. This means an analysis needs to be done several times until the settings are optimal applied.
The mentioned issue that there is no documentation makes it difficult to apply all the settings for the analysis. For instance it is not possible to apply the settings below the threshold area, because no one knew what its purpose is. By contacting Adrie Raijmaaker who developed the software it was possible to get an explanation of most of the settings but he did not know the purpose of some settings either. But he was able to add a few functions to the program.
The analysis of the total distance and "mean velocity" in first table "tracks" and the distance, velocity, acceleration and area in the second table "subtracks" of the analysis text file were added by the software developer. The columns for total distance in the first table and the distance and velocity of the second table were given in the first place but did not contain any information. With the new added information's more detailed analysis can be done with Matlab.
