Thread Subject: fruit features extraction

Hi, I'm working on a project to classify fruit based on its category. Is anyone can tell me how to extract fruit features such as colors, shape, size or weight using matlab?

Assume you have an RGB image, you will find Red, Green and Blue values are stored in separate matrices, i.e. Image(:,:,1), Image(:,:,2), and Image(:,:,3). You can extract color by looking at these color spaces.

Shape extraction is basically an image segmentation problem. You might want to first take advantage of color information and separate the fruits by colors. Then you might want to threshold your image (graythresh, im2bw), or use edge detection algorithms (edge). Finally you might need to label each components (bwlabel, regionprops).

Size is easy. Once you find the shape, just count how many pixels are inside the shape.

type help on rgb2gray, im2bw, graythresh, edge, bwlabel, regionprops...

As to weight ... I am not sure if it's relevant.

"Fatma " <fit@itc.udm.edu.my> wrote in message <heg4id$cgq$1@fred.mathworks.com>...
> Hi, I'm working on a project to classify fruit based on its category. Is anyone can tell me how to extract fruit features such as colors, shape, size or weight using matlab?

On Nov 24, 3:13 am, "Fatma " <f...@itc.udm.edu.my> wrote:
> Hi, I'm working on a project to classify fruit based on its category. Is anyone can tell me how to extract fruit features such as colors, shape, size or weight using matlab?

-----------------------------------------------------------------------------------------
Someone asked about this last April (http://groups.google.com/group/
comp.soft-sys.matlab/browse_frm/thread/f281f673b7a2c1c2/
ff3c1b3d80debbe9?hl=en&lnk=gst&q=fruit#ff3c1b3d80debbe9)

Basically you have to segment your image based on color, edges, shape,
or whatever you can do separate each fruit out individually. Then
create your feature vector of measurements (size, shape, color,
roundness, etc.) and compare it to feature vectors of "known" fruits.

I can't say much more unless you post your images somewhere.

"Clare " <flyingclare@gmail.com> wrote in message <heg5oj$p03$1@fred.mathworks.com>...
> Assume you have an RGB image, you will find Red, Green and Blue values are stored in separate matrices, i.e. Image(:,:,1), Image(:,:,2), and Image(:,:,3). You can extract color by looking at these color spaces.
>
> Shape extraction is basically an image segmentation problem. You might want to first take advantage of color information and separate the fruits by colors. Then you might want to threshold your image (graythresh, im2bw), or use edge detection algorithms (edge). Finally you might need to label each components (bwlabel, regionprops).
>
> Size is easy. Once you find the shape, just count how many pixels are inside the shape.
>
> type help on rgb2gray, im2bw, graythresh, edge, bwlabel, regionprops...
>
> As to weight ... I am not sure if it's relevant.
>
> "Fatma " <fit@itc.udm.edu.my> wrote in message <heg4id$cgq$1@fred.mathworks.com>...
> > Hi, I'm working on a project to classify fruit based on its category. Is anyone can tell me how to extract fruit features such as colors, shape, size or weight using matlab?

Thanks a lot Claire. That will be very helpful for a beginner like me



"Clare " <flyingclare@gmail.com> wrote in message <heg5oj$p03$1@fred.mathworks.com>...
> Assume you have an RGB image, you will find Red, Green and Blue values are stored in separate matrices, i.e. Image(:,:,1), Image(:,:,2), and Image(:,:,3). You can extract color by looking at these color spaces.
>
> Shape extraction is basically an image segmentation problem. You might want to first take advantage of color information and separate the fruits by colors. Then you might want to threshold your image (graythresh, im2bw), or use edge detection algorithms (edge). Finally you might need to label each components (bwlabel, regionprops).
>
> Size is easy. Once you find the shape, just count how many pixels are inside the shape.
>
> type help on rgb2gray, im2bw, graythresh, edge, bwlabel, regionprops...
>
> As to weight ... I am not sure if it's relevant.
>
> "Fatma " <fit@itc.udm.edu.my> wrote in message <heg4id$cgq$1@fred.mathworks.com>...
> > Hi, I'm working on a project to classify fruit based on its category. Is anyone can tell me how to extract fruit features such as colors, shape, size or weight using matlab?

Thanks, since I am zero, it will be quite a while for me to catch all those thing

ImageAnalyst <imageanalyst@mailinator.com> wrote in message <3352d54b-f7b8-4056-a303-a4fe5d9adb68@k9g2000vbl.googlegroups.com>...
> On Nov 24, 3:13?am, "Fatma " <f...@itc.udm.edu.my> wrote:
> > Hi, I'm working on a project to classify fruit based on its category. Is anyone can tell me how to extract fruit features such as colors, shape, size or weight using matlab?
>
> -----------------------------------------------------------------------------------------
> Someone asked about this last April (http://groups.google.com/group/
> comp.soft-sys.matlab/browse_frm/thread/f281f673b7a2c1c2/
> ff3c1b3d80debbe9?hl=en&lnk=gst&q=fruit#ff3c1b3d80debbe9)
>
> Basically you have to segment your image based on color, edges, shape,
> or whatever you can do separate each fruit out individually. Then
> create your feature vector of measurements (size, shape, color,
> roundness, etc.) and compare it to feature vectors of "known" fruits.
>
> I can't say much more unless you post your images somewhere.