can images be given as inputs to neural network GUI toolbox?

i'm doing a project on road extraction using ANN. i have a problem in training the network using input image and manually extracted image..pls help...

Answer 1

[Greg Heath]

1. Use feature extraction to reduce the number of pixels

2. Columnize the resulting image into a vector using (:)

Hope this helps.

[shakthi g d]

thank you sir...can u send me a code for feature extraction of road from background image? and can you tell me if its better to use nntool gui interface or to use coding for creating and training the network...

Answer 2

[Image Analyst]

See Sections 15 and 22 here: http://iris.usc.edu/Vision-Notes/bibliography/contents.html You should know how people are successfully detecting roads in aerial images, or following roads by autonomous vehicles (not sure which one you're doing). They may or may not be using neural networks.

[shakthi g d]

thank you sir...really inspiring....we are trying to do road detection....

[shakthi g d]

i saw your image segmentation and a few other demo codes...good job...can you please help me out to get a code of feature extraction of road from background?