Query regarding Computer Vision

Can I detect the position of still image in different state ?????Is there any example ???

[Walter Roberson]

Please explain your question more?

[Mrinmoyee Bhattacharya]

I want to detect the the position of cyclonic image in different states...can I do it using computer vision or machine learninr or anything ?

[Walter Roberson]

Do you mean that you have a number of sample images of what a cyclone looks like in different states, and you want to be able to take an input image of a cyclone and classify it according to which of the states it is in?

[Mrinmoyee Bhattacharya]

yes

[Image Analyst]

Allow us to help you and answer by reading this link and taking the corrective actions described in there. Otherwise we're working "blind". You have not shown us any image (like people normally do when asking for image processing advice), nor have you decribed exactly what "position" means to you (centroid, boundary, all pixels, feret diameters, or ???). And what does "state" mean? Visible size? Category based on wind speeds near eye wall? Do you have a wind speed image, or just a reflectance, visible light image?

[Mrinmoyee Bhattacharya]

Basically i want to say that like object tracking, in object tracking object is movable but to me it is still image. I have some wind speed image. Suppose cyclone is formed in bay of bengal on 2nd october and on 3rd october it turns around arabian sea. I just want to track the position of those image.

[Image Analyst]

Yep, okay. Thanks for your partial answers. I'll await your uploads and complete explanations. I'll check back later.

[Mrinmoyee Bhattacharya]

In this attachment i have given some wind speed image. Yellow points represents wind speed. After seeing you will understand what i want to say. i just want to track the position of wind speed using computer vision or machine learning or anything, which will say the position of the wind speed

[Walter Roberson]

You talk about having "still images". But note that all that a "movie" is, is a series of still images in time lapse.

Answer 1

[Image Analyst]

There is not much that can be done with your PDF document of map line drawings that you uploaded. It's pretty much useless. But you said you have images of wind speed, like where each pixel represents some number of km per hour at that location. So all you have to do, if you want the centroid location, is to threshold the windspeed image and find the centroid

binaryImage = windSpeedImage > someThreshold; % For example someThreshold = 150 km per hour or whatever.
binaryImage = bwareafilt(binaryImage, 1); % Take just the largest blob.
props = regionprops(binaryImage, 'WeightedCentroid', 'Centroid');

By computing the distance between centroids at those different time point images, you can find out how fast the centroid moves (average velocity) from one time to the other.

See My File Exchange for a full tutorial on image segmentation.