How to convert 1.35678e-01 to numerical digits in matalb?

Hello,

I am doing image processing and after segmentation i extract few Features like GLCM, entropy,Area etc. All these answers come as 1.3456e-01, 3.564676e+01 etc. . Is there any way to convert this to normal numerical numbers? I use Matlab 2014a. Please help me with respect to this.Thank you

[Joel Handy]

What do you mean by "normal numerical numbers"? Those are normal numbers. Are those values coming back as strings you need to convert to numbers? If thats the case str2num or str2double will convert them. Otherwise its just display formatting. Try "format long g" to change how the command window displays numbers.

[Ramanathan Anandarama Krishnan]

Hello Joel Handy,

For example i use the following commands:

clc;
clear all;
close all;
a=imread('coins.png');
glcm=graycomatrix(a);
stats=graycoprops(a);

But i get output as follows:

Contrast: 1.2782e+04

Correlation: -0.0012

Energy: 1.7541e-05

Homogeneity: 0.0327

As u can see contrast and energy are in scientific notation. I need it in normal whole numbers like 0.005, 12345, 89898 etc. Please help me in this.

Answer 1

[Adam Danz]

" I need it in normal whole numbers like 0.005, 12345, 89898 etc"

Setting the format to longg (not long!) or shortg will solve the problem for the 'Contrast' value but not the 'Energy' value.

format longg
stats = 

  struct with fields:

       Contrast: 12781.9263669137
    Correlation: -0.00117293439093276
         Energy: 1.75410617429713e-05
    Homogeneity: 0.0326985233071923
    
    
format shortg
stats = 

  struct with fields:

       Contrast: 12782
    Correlation: -0.0011729
         Energy: 1.7541e-05
    Homogeneity: 0.032699

How will you be using the numbers? If you're just passing them on to another function this change in format does nothing. If you're printing them, you can convert them to strings like this

sprintf('%.6f',stats.Energy)
ans =
    '0.000018'

[Ramanathan Anandarama Krishnan]

Hi adam,

Thanks for your code and it's really working for the image. These are features i am extracting from medical images and further it will be used for classification. Thus i think format longg or shortg will not be the best as it is working only in command window and thus not make a difference in variable stored as stats. (Variable of contrast still remains like this 1.278192636691366e+04 in editor but in command window it is 12782)....

1.) Can i proceed with numbers like 1.0023e+05 for classification?

2.) If not, how to change it in variable instead of command window.

And why the code is not working for energy.

[Adam Danz]

"These are features i am extracting from medical images and further it will be used for classification. "

It sounds like you don't need to change the format of the number at all. The format just changes the display. 0.000007 and 7e-06 are the same number in matlab and your classifier will behave the same with either input.

"Can i proceed with numbers like 1.0023e+05 for classification?"

A definite yes.

" If not, how to change it in variable instead of command window."

There is no way. You could always round numbers but then you lose precision and your classfier doesn't benefit from that at all. In fact, it's a bad idea.

"why the code is not working for energy"

The first significant digit needs to appear within the first 4 decimal places with shortg and longg formats. Otherwise scientific notation is used.

[Ramanathan Anandarama Krishnan]

Dear Adam,

Thank you soo much for such crystal clear explanation. I will proceed with classification as u said.

[Adam Danz]

Glad I could help!

[Walter Roberson]

In the variable editor, at the top of the window for each variable, there is a drop box that permits you to change the display (not storage) format for the variable. There is also a Preference for the default to use.

[Ramanathan Anandarama Krishnan]

Hi Adams/Walter, I am manually copy pasting my features to Excel.However except area feature which comes as 4 digit number like 4000 etc all the other features are in scientific notation i mean 3.34****E+02.Is this fine for classification or all features must be in 1 type only? Please reply.

[Walter Roberson]

They will all become datatype double when you load them from excel.

[Ramanathan Anandarama Krishnan]

Then I have no problem. Thanks a lot Walter