Want to use matlab to distinguish the four pictures, walls, trees, utility poles and power line entirely, I now want to extract features, and then input into the support vector machine (SVM) to classify, before calculating the gray level co-occurrence matrix, extracts the four texture characteristic value, but the effect is not obvious,

So want to ask bosses are there other ways to extract other features? Thank you for your attention!

CodePudding user response:

Data continuous wall
Forest data discrete
Narrow pole figures, concentrated
Power line is bad, make the signal is too weak to be

Test can see the effect first take the field lines

CodePudding user response:

Directly on CNN, characteristics of the project is too complicated

CodePudding user response:

The poker named, mark

CodePudding user response:

With analysis of characteristics of image binary

CodePudding user response:

If the sample lot, CNN, sample is not much, their characteristics, and then support vector machine (SVM) and so on

CodePudding user response:

Consider it as a voice, extract the characteristics of the audio?

CodePudding user response:

Spectrograms (spectrogram)

Python painting speech spectrum

CodePudding user response:

The typical sound file,,,,,,,,,,