draco1023 / aforge

Automatically exported from code.google.com/p/aforge
Other
0 stars 0 forks source link

Unable to detect object after changing the angle of abject which was different while training object. (getting diff. dimentions) #150

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago
What steps will reproduce the problem?
1. Train the object with blob detection algorithm & get its dimentions height 
width.

            int blobWidth, blobHeight;
            arr1[i] = new Object[] { "", "", 0, 0, 0, 0, NOTDETECTED };
            ApplyFilter(new ExtractBiggestBlob());
            string TrainingBigBlobPath = Application.StartupPath + "\\training\\biggest_blob.jpg";
            image.Save(TrainingBigBlobPath);
            blobHeight = image.Height;
            blobWidth = image.Width;

2. In inspection mode we have to detect the object from various objects in 
image. Now if i change the angle of object, it wont detect as the blob 
extractor algorithm returns the detected blob's height, width which is 
different that what i have while training that object. 

3. In this case it wont detect the object in image, i have to again change the 
position(angle) of object to detect.

What is the expected output? What do you see instead?

I want my algorithm to detect the same object from different angles. I should 
be able to recognize the object from various angles. I am attaching images for 
your reference

Please use labels and text to provide additional information.

Original issue reported on code.google.com by wadekarashish@gmail.com on 2 Sep 2010 at 12:27

Attachments:

GoogleCodeExporter commented 9 years ago
Most probably you don’t understand the idea of blob detection algorithm at 
all.

BlobCounter just finds separate blobs. This is it. There is no any type of 
training relatad to class. It has no knowledge about what you’ve found in any 
previous image and I am not sure where did you get this expectation from.

By the way. There is documentation available – check it.

Original comment by andrew.k...@gmail.com on 2 Sep 2010 at 12:42

GoogleCodeExporter commented 9 years ago
Actually I am new to image processing. 
Could you please tell me is there any way i can make this work..

I want to detect the object from different angles.

Original comment by wadekarashish@gmail.com on 2 Sep 2010 at 12:49

GoogleCodeExporter commented 9 years ago
You may need to use rotation invariant template matching.

Check things like SIFT and SURF:
http://en.wikipedia.org/wiki/Scale-invariant_feature_transform
http://en.wikipedia.org/wiki/SURF

Original comment by andrew.k...@gmail.com on 2 Sep 2010 at 2:01

GoogleCodeExporter commented 9 years ago
Thank You very Much... !!!
I'll give a try.... !!!

Once again Thanks For Your Time... !!!

Original comment by wadekarashish@gmail.com on 2 Sep 2010 at 3:05