Skip to content

Commit

Permalink
Updated both panorama and plate-detection READMEs
Browse files Browse the repository at this point in the history
  • Loading branch information
bvanelli committed May 20, 2018
1 parent 3a5548f commit 926847a
Show file tree
Hide file tree
Showing 2 changed files with 27 additions and 1 deletion.
2 changes: 1 addition & 1 deletion panorama/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -110,7 +110,7 @@ As expected, Photoshop renders the best results for the image stitching.

## I want to run it!

First, [download and install the toolbox](http://petercorke.com/wordpress/toolboxes/machine-vision-toolbox#Downloading_the_Toolbox) (for no particular reason, all the strings in the toolbox use the wrong quotes that does not work in 2016b and below. If you are facing issues, search and replace all backticks ``(`)`` with `'`). Just replace the photos on the `dataset` folder with your own, or change the path in the second line of the script.
First, [download and install the toolbox](http://petercorke.com/wordpress/toolboxes/machine-vision-toolbox#Downloading_the_Toolbox) (for no particular reason, all the strings in the toolbox use the wrong quotes that does not work in 2016b and below. If you are facing issues, search and replace all double-quotes `"` with single quotes `'`). Just replace the photos on the `dataset` folder with your own, or change the path in the second line of the script.

```matlab
images = iread('YOUR FOLDER HERE/*.JPG','double');
Expand Down
26 changes: 26 additions & 0 deletions plate-detection/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -67,3 +67,29 @@ The ZNCC is a pretty robust algorithm for template matching because it's invaria
Using the first character of the plate used as an example, the letter B, we can obtain the results matching with all the letters available on the template. Notice that the B gives us the best result, but other round characters like D or S give high values, meaning that trying to match low quality plates (like in pictures taken from far away) might be a problem.

<img src="https://user-images.githubusercontent.com/8211602/40282847-6d286e74-5c4b-11e8-84b4-ed414029d6ae.png" width="50%">

## I want to run it!

First, [download and install the toolbox](http://petercorke.com/wordpress/toolboxes/machine-vision-toolbox#Downloading_the_Toolbox) (for no particular reason, all the strings in the toolbox use the wrong quotes that does not work in 2016b and below. If you are facing issues, search and replace all double-quotes `"` with single quotes `'`). First, import your image using your favorite Computer Vision Toolbox in grayscale and double precision.

```matlab
plate = iread('YOUR IMAGE HERE', 'double', 'grey');
```

You also need to load the template for template matching (we include the mandatory template in the 'fonte' folder):

```matlab
template = load_font('fonte/letras.png', 'fonte/numeros.png');
```

Now, try identifying the plate:

```matlab
s1 = get_plate(plate, template)
```

Alternatively, you can identify the state-city:

```matlab
h1 = get_plate_header(plate, template)
```

0 comments on commit 926847a

Please sign in to comment.