Forum Discussion

Colin_McCrae's avatar
Colin_McCrae
Community Hero
8 years ago
Solved

Image Configuration properties?

I've been messing about with OCR to try and at least do something with a system we use which simply presents the user with a large Canvas. The Canvas is actual lots of buttons (all of which are fully...
  • HKosova's avatar
    HKosova
    8 years ago

    Hi Colin,

     

    Here's how to specify the color reduction for GIF:

    // JScript
function Test()
{
  var img = Utils.Picture;
  img.LoadFromFile("C:\\MyImage.bmp");

  var gifConfig = img.CreatePictureConfiguration("GIF");
  gifConfig.ColorReduction = "Windows system (20 colors)";

  img.SaveToFile("C:\\MyImage.gif", gifConfig);
}

    We'll look at getting some more examples in the docs. Thanks for the feedback!

HKosova's avatar
HKosova
SmartBear Alumni (Retired)

Hi Colin,

 

Here's how to specify the color reduction for GIF:

// JScript
function Test()
{
  var img = Utils.Picture;
  img.LoadFromFile("C:\\MyImage.bmp");

  var gifConfig = img.CreatePictureConfiguration("GIF");
  gifConfig.ColorReduction = "Windows system (20 colors)";

  img.SaveToFile("C:\\MyImage.gif", gifConfig);
}

We'll look at getting some more examples in the docs. Thanks for the feedback!

Colin_McCrae's avatar
Colin_McCrae
Community Hero
8 years ago

Thanks Helen.

 

One of the dev guys here suggested after I posted it probably worked this way.

 

So as much due to my lack of dev background. Although I do still think some better examples would help. The way I was reading it I thought I could pass those properties in as arguments. Obviously not!

 

It seems to be helping with the OCR. I have it working pretty well. But due to the massive potential variance in size, position, and colour, it helps if I can simplify and standardize the image as far as possible before running the OCR over it. I'm having much better success once they are converted to monochrome/black and white. Although I'm still trying to flatten out a little more of the dithering/aliasing on the image font as it causes some letters to appear joined together, which the OCR completely fails to spot. Not surprisingly. Even if I can get it around 90% accurate, that would still be a win in terms of manual man hours spent testing these things.