You are on page 1of 25

IMAGE PROCESSING APLLIED TO AGROINDUSTRY

OBJECTIIVES
Research techniques applied to agro industrial sector,

supported by theories of artificial vision. The development of technology oriented to an improvement in the selection of fruits, in order to keep the required standards of quality and add value to the products of the region. Apply this techniques using MATLAB IMAGE TOOLBOX. Achieve communication between the camera and MATLAB in order to capture images and process its.

BACKGROUND
Digital image:
It is a representation of a bi dimensional image using a

finite number of point, called pels or pixeles, which are represented by one or more numeric values. For monochromatic image, just one value represents the pixel intensity (0 255). For color image, uses 3 values, which represents Red, Green and Blue colors (RGB) components.

Digital image processing:


It is the manipulation of images, supported by a

computer, generally made in an automatic way and those are based in algorithm designed carefully. There are three levels of operation in image processing:

Low level: Noise reduction, contrast enhancement. Inputs and outputs are images. Medium level: Edges, perimeter, areas, etc. Of images. High level: analysis and interpretation of the content of the scenes

Color Analysis
Color images are represented by RGB that indicates a

MxNx3 color pixels matrix. They can be seen as a stack of three grey scale images. RGB models are based on Cartesian coordinates systems whose axes represent Red, Green and Blue colors.

Digital images representation Bitmap

Uses one or more bi dimensional pixel matrix Uses drawing commands to represent a image

Vector

Binary Image

2D pixel matrix. <0 1>. 0 means totally black and 1 means totally white.
2D pixel matrix. 8 bits per pixel. [0 255]. 0 means totally black. 255 means totally white. Values between 0 and 255 mean lot of grey tones.

Gray Scale image

Color Image

RGB: for each color channel (R G B), corresponds a 2D matrix, which each element keeps a 8 bit value, that indicates the quantity of Red, Green and Blue in that determined point, in a [0 255] scale. Indexed color: for antique hardware that isnt able to show 16 millions of colors simultaneously.

Algorithms used Size algorithm


Calculates the width and height of the fruit. We proceed binarizing the image in order to find the height and width counting valid pixels in rows and columns.

Shape algorithm

Qualify the roundness of the fruit, independently of the physics features.

Shape algorithm flow chart

Color algorithm

Set up an histogram of the R G B components to analyze the quality of the fruit

Color algorithm flow chrat

Testing
At first we proceed to obtain a controlled environment,

to isolate the fruit from external noise. In addition, we set up a light source in order to have a good image. The ground might be white color

Distance testing:

The camera is set up at 52.5 cm high from ground. Capture image.

1 2 3 4 5 6 7 8

Medida real [cm] Medida computarizada [pxeles] Relacin Permetro Altura Ancho Permetro Altura Ancho Permetro Altura 33 11,855 8,687 1627,6 564 400 49,32121 47,57486 32,8 11,237 8,73 1584,1 543 400 48,29573 48,32251 34 12,826 9,026 1704,5 609 408 50,13235 47,48168 31,9 11,307 8,288 1540 526 385 48,27586 46,51985 37 13,929 9,529 1833,5 654 449 49,55405 46,9524 26 9,004 7,412 1270 416 344 48,84615 46,20169 31,5 11,117 8,641 1528,6 516 398 48,52698 46,4154 32,5 11,625 8,622 1608,1 552 403 49,48 47,48387 PROMEDIO 49,08368 47,21704

PROMEDIO GENERAL Ancho 46,04582 45,81901 45,20275 46,4527 47,11932 46,41123 46,05948 46,7409 46,23535

47,21703965

Relacin Permetro
50.5

50

49.5

49

y = -0.0037x + 49.071 R = 0.0002

48.5

48

Relacin Altura
48.5

48

47.5

47

46.5

y = -0.1617x + 47.847 R = 0.3012

46

45.5

45 1 2 3 4 5 6 7 8

Relacin Ancho
47.5

47 y = 0.1233x + 45.676 R = 0.2638 46.5

46

45.5

45

44.5

44

Color testing

Configuracin de la cmara
imaqhwinfo('winvideo')
ans =

AdaptorDllName: [1x81 char] AdaptorDllVersion: '4.5 (R2013a)' AdaptorName: 'winvideo' DeviceIDs: {[1] [2]} DeviceInfo: [1x2 struct]

info=imaqhwinfo('winvideo',1) info =

DefaultFormat: 'M420_1280x720' DeviceFileSupported: 0 DeviceName: 'Microsoft LifeCam Studio' DeviceID: 1 VideoInputConstructor: 'videoinput('winvideo', 1)' VideoDeviceConstructor: 'imaq.VideoDevice('winvideo',

1)'

SupportedFormats: {1x36 cell}

>> info.SupportedFormats ans = Columns 1 through 5 'M420_1280x720' 'M420_160x120' 'M420_176x144' 'M420_1920x1080' 'M420_320x240'

Columns 6 through 10 'M420_352x288' 'M420_424x240' 'M420_640x360' 'M420_640x480' 'M420_800x448'

Columns 11 through 15 'M420_800x600' 'M420_960x544' Columns 16 through 20 'MJPG_1920x1080' 'MJPG_320x240' 'MJPG_352x288' 'MJPG_432x240' 'MJPG_640x360' 'MJPG_1280x720' 'MJPG_160x120' 'MJPG_176x144'

Columns 21 through 25 'MJPG_640x480' 'MJPG_800x448' 'MJPG_800x600' 'MJPG_960x544' 'YUY2_1280x720'

Columns 26 through 30

'YUY2_160x120'

'YUY2_176x144' 'YUY2_1920x1080'

'YUY2_320x240'

'YUY2_352x288'

Columns 31 through 35 'YUY2_424x240' Column 36 'YUY2_960x544' 'YUY2_640x360' 'YUY2_640x480' 'YUY2_800x448' 'YUY2_800x600'

You might also like