Steve on Image Processing

February 16th, 2006

I learned something interesting this week in conversations with MATLAB users. By e-mail and phone I talked with two people who have been reading posts in this blog. Both have the Image Processing Toolbox. One I know personally; he is an image processing expert and has used MATLAB and the toolbox for many years.

As it turns out, neither user knew about the toolbox functions that I mentioned in my previous spatial transformations post. I admit that this surprised me a bit. These functions were added to the product almost five years ago, when Image Processing Toolbox version 3 was released.

We software developers know in our heads how long it can be before users learn and start to use new features, but it can be an abstract concept that's hard to remember. Also, our mental release timeline is shifted compared to our users' timeline. When a release ships, its new features may have been largely completed by developers some months before; those developers have already moved on to something else. And it may be months longer before the new release is actually installed and available at many customer sites.

I really shouldn't have been surprised, I suppose. I'm sure there are lots of useful features I don't know about in software that I use every day.

20:58 UTC | Posted in Uncategorized | Permalink |

You can follow any responses to this entry through the RSS 2.0 feed. You can skip to the end and leave a response. Pinging is currently not allowed.

5 Responses to “Resetting my expectations: What do users know about new product features?”

1. Richard replied on February 21st, 2006 at 17:43 UTC :
   

   Hi Steve

   I am one of those people who wasn’t aware of the tform functionality, despite having used the image processing toolbox for the last year and a half for some of my PhD work.

   I suppose it’s because most of the time the transformations I work with are at worst projective, so I only ever need to use constant matrices (and their inverses). Because I tend to work with points extracted from images, rather than images themselves, points are easily transformed by simple matrix multiplication. However, from reading the documentation, imtransform looks like a very useful function!

2. Greg replied on February 24th, 2006 at 03:53 UTC :
   

   Hi Steve,
   I’m another of those who didn’t follow the tform introduction. However, I’m casting around in the documentation now because I’d like to do polar and back to cartesian transformations of images. Is there something that will do non-affine transforms? Any pointers? Google is letting me down…

   Cheers,
   greg

   PS more blogs from TMW would be great.

3. Steve replied on March 6th, 2006 at 08:33 UTC :
   

   Greg - imtransform can handle many types of spatial transformations. However, it’s meant for transforming from one Cartesian grid to another. If you have image samples in polar coordinates and want to create an image on a Cartesian grid, I suggest you look at griddata.

   If these first few blogs are successful, then I predict there will be more.

4. jani replied on March 30th, 2006 at 01:43 UTC :
   

   please tell me how to do multiplication in image processing without using im2double conversion ?

5. Steve replied on March 31st, 2006 at 08:09 UTC :
   

   Jani - I assume you are asking how to multiply an integer array by some floating-point scalar. In MATLAB 7, this works fine; you don’t need to convert to double first. In earlier versions, use the Image Processing Toolbox function immultiply.

Leave a Reply

<pre class="code">
a = magic(3);
sum(a)
</pre>

These postings are the author's and don't necessarily represent the opinions of The MathWorks.