OCR Images (No No, Right)?

[TheMcD]

I need to make sure I have things correctly understood.

(I don't own MS at the moment.)

I cannot OCR an image with the default MS, as in:

http://screencast.com/t/JZabPDjhy

The image is originally a PNG file.

-----

The only way that seems possible is using: Textract, as mentioned here:

https://www.mjtnet.com/blog/2006/06/06/ ... ical-text/

Thanks

[Phil Pendlebury]

...Or you could do a screen capture of the numbers and a little logic juggling. It wouldn't be too difficult, as it is only numbers you are after.

[TheMcD]

...Or you could do a screen capture of the numbers and a little logic juggling. It wouldn't be too difficult, as it is only numbers you are after.

Would you please elaborate?

Are you saying something like looking for an image match for each number 0-9 until the whole number is generated, then move on to the next number?

Example: RSI (from above image), Find a 5, then a 4, then a ".", then a 0, then an 8?

Thanks

[Phil Pendlebury]

Yup pretty much. Have a variable for each individual image of 1 - 9.

Do the math / logic to convert the variables into numbers.

It would involve quite a bit of code but it is certainly do-able.

So for example: Finding 17

Your script would search for any number 1-9, if found would then write to avar and look next to it for another number etc. If no further number is found, it would stop and then convert the variable to a number.

It is quite easy to split the screen to look at certain areas and also to look at an area any number of pixels next to the previous area.

Having said that, if you have no scripting skill at all this may be a bit of task.

[TheMcD]

To add to the conversation, here is a non-marked up image:

http://stockcharts.com/c-sc/sc?s=AAPL&p ... 3023542151

[Phil Pendlebury]

I would suggest you get the trial of Msched and start to learn how it works, or consider employing someone to write the script for you.

As I mentioned above this is not a beginners task, and although you will get a lot of help here, you will not get the entire script written for you, you have to show a bit of interest in at least starting to write it yourself.

Maybe Marcus can add to this?

[TheMcD]

Thanks to Phil for his suggestion.

Marcus, anyone else?

Surely this can't be the only way... ?

[Marcus Tettmar]

There are other OCR engines. As well as Textract, there is the free Microsoft MODI which is part of Office and can be used from Macro Scheduler VBScript, and a number of more powerful third party commercial solutions which usually come with an SDK you can then use with Macro Scheduler.

[mightycpa]

Just for fun, I tried it, and by jiggling a couple of settings, it seems to work for me.

Looking at this (can't embed the image for some reason)

https://photos.google.com/share/AF1QipO ... 11bEN2bFRn

I got this:

AAPL (Daily) 114.81
MA(S0) 125.33

MA(200) 120.05

Volume 25,185,958

[TheMcD]

mightycap,

What did you use for the OCR?

[mightycpa]

I used Tesseract, but I'll tell you right now, it won't work reliably "right out of the box"... like I said, you'll have to jiggle some of your settings. If you look carefully at what I got, you'll see that the 50-day moving average is described as MA(S as in Sam O). So I'll probably have to futz around with it a little more, but it's 90% right now.

I could do it for you next week. Try it for yourself, and if you can't make it work, let me know, we'll work something out.