|
|
Hello Gentlemen,
I have a less elementary question this time.
I have application that runs both full screen and in a window. Text appears on the screen. I want to take each frame of this application or perhaps a new snap shot every few seconds, and then OCR the frame to extract the text. I'd like to do this in c# but I do not know of a way to get the handler for this application and extract and store it's pixel information. This is probably going to be really resource intensive and I'd like to keep it as fast as possible.
What i'm basically looking for is a starting point. How can I get that pixel data from the external application into a format I can edit. I only need to monitor a certain section of the screen of this application. Is there a FCL class that I can look at to get me on my way?
Thank you.
|
|
|
|
|
If you're in C# 2.0, then you can use a method on the Graphics object to copy data from the screen to a bitmap. THe OCR stuff is a little more difficult. If you know what the app is and can get the window handle of the controls displaying text, GetWindowText is probably a better option ( that's a C API ).
Christian Graus - Microsoft MVP - C++
Metal Musings - Rex and my new metal blog
"I am working on a project that will convert a FORTRAN code to corresponding C++ code.I am not aware of FORTRAN syntax" ( spotted in the C++/CLI forum )
|
|
|
|
|
The application i'm trying to get the text out of is a video game that can run in full screen or in window. Most likely the test being displayed is through some kind of proprietary direct x game engine that I have no control over. The only thing I can do is monitor the games' frames and pick out of them the information I want to perform OCR on. I like the idea about getting the window handle of the app. If I get the window handle of the app, how can I gain access to it's pixel data, or does that just intrinsiclly come with it? I'm a bit confused when you mention the "controls displaying text" because, as far as I know there are no controls (at least as far as I know of their definition goes in C#, ex, button, textbox, etc. I could easily be confused about the definition of controls, so don't rule that out.
I was hoping to avoid having to copy data from the screen to bitmap as that could take a very very large amount of data and cpu cycles if i'm doing that 30 frames per second. But the nice thing is that the text isn't constantly altering data and just kind of sits there until it scrolls away.I probably just need to capture 1 frame every second, but will the OCR (which can be pretty heavy itself) be able to compute and recognize that text before the next frame is pulled? It probably doesn't even need to be that quick as I could create an ocr buffer that checks to see if it already got that text and I can put a time in milliseconds next to that text to rule out the OCR not reporting duplicates.
This is just a hobby project by the way, ambitious as it may sound.
|
|
|
|
|
There's a problem with this. You can probably get the image from the screen, but only probably. There's no guarantee that you will be able to do this. Since this is a game, the screen updates will be happening VERY rapidly. This causes you two problems.
Since your talking about scrolling text, there is an issue of shearing. This is when you start to grab a shot of the area of the screen and it changes DURING the grab. This can make the OCR part fail quite easily.
The second problem IS the OCR part. This will be a very CPU heavy process that can hog the CPU and slow down the game performance. You definitly won't be getting this at 30 frames a second. Frame a second will be far more likely and this will leave more CPU time to the game itself.
|
|
|
|
|
I'm glad you brought shearing to my attention, i was not aware of it.
As far as the OCR goes, I would have a very fast machine dedicated to doing this one function of monitoring text.
I do not need to get every single frame. The text is only entered by humans and in a sense can be logged graphically. I could easily limit things to grabbing 1 frame every 60 or even every 120.
What I do need to be able to do is to get past the shearing.
The only thing i can think of to get past this is to somehow preempt the rendering pipeline of the graphics card and capture a complete frame buffer. If i capture a complete buffer then I'm pretty much assured that its not half of the new frame overwriting half of the old frame. If the screen was updateing during that z-buffer flip then i can just have the OCR drop everything that it doesn't pick up. If one line of the text has an unrecognizeable character, I just drop the posting of that line and wait until the text scroll animation has been completed. Remember, none of the frames have to actually be displayed to screen, which saves some cycles. It just has to be captured into a readable format. Hopefully there is another way without haveing to use the screen output as a proxy to the z-buffer.
Anyone know how I can get the z-buffer information? Is there an API method that lets me pipe the graphic output of the game into another app for processing?
|
|
|
|
|
Kataric wrote: Is there an API method that lets me pipe the graphic output of the game into another app for processing?
Is there an API to do this?? No.
Kataric wrote: The only thing i can think of to get past this is to somehow preempt the rendering pipeline of the graphics card and capture a complete frame buffer
Good Luck with that. The only way you could even come close to doing this is if you injected code into the game process. This is something that cannot be done with managed code. You'll have to write this part in unmanaged C++ to support the exports and static linking required to do DLL injection.
|
|
|
|
|
I'd like to use c# by the way to do this. If no managed code can handle it, I'll go back to c++.
|
|
|
|
|
I have added a large number of jpg files to my project's resource file of a windows application. How do I programmatically get to these images in C# code. I need a code sample please.
joe
|
|
|
|
|
Your_Name_Space.Properties.Resources.Your_Resourse;
Smile: A curve that can set a lot of things straight!
(\ /)
(O.o)
(><)
|
|
|
|
|
I am having issues with the NewWindow event of the WebBrowser control in .NET 2.0. If I click on any links with a target of _blank I am expecting it to open my new form with the browser control in it. What happens is that it opens my form and then Internet Explorer 7 on top of that with the contents I was expecting to appear in my form. Is there a way for me to stop it from opening IE? Has anyone had this issue before?
-Mike S.
|
|
|
|
|
You can suppress popups, but any popups will open IE ( and IE is what you've got open all along, it has no idea it's in your form )
Christian Graus - Microsoft MVP - C++
Metal Musings - Rex and my new metal blog
"I am working on a project that will convert a FORTRAN code to corresponding C++ code.I am not aware of FORTRAN syntax" ( spotted in the C++/CLI forum )
|
|
|
|
|
That is still confusing to me. It is not really your typical popup because it was a link that I clicked on. I wasn't something opened by script code in the HTML code. So if I am controlling the event and telling it to open a new form with a browser control in it why would it override that and open a second window on top of that? I did see in my Google searches that people referenced something called NewWindow2 and NewWindow3 but that doesn't seem to be available for trapping under the web browser control.
|
|
|
|
|
I should mention that this worked correctly under the old ActiveX browser that we were using before.
|
|
|
|
|
The NewWindow event fires AFTER you click on a link that opens a new window (OR) use the "Open in a New Window" context menu (AND) BEFORE the new window actually opens. To cut the crap, this wont open a new Form with another webbrowser control on it. It'll only launch a new IE window.
If you want to open a new Windows Form, you've to put in your own code that opens a new Form with another web browser control and cancel the event so that IE is not launched.
SG
|
|
|
|
|
Once I was able to keep in mind that another event was firing to create the IE instance then I was able to cancel it and figure out what my other issues were. Thank you very much!
|
|
|
|
|
Hi,
I am new to C# Xml programming. Basically, I am trying to save a XML file to a database. I found that if I just take the below string and try to save it a table which contain a XML datatype column, it creates a SQL Exception:
SqlException: XML parsing: line 1, character 38, unable to switch the encoding
String Test_PSQ1 = @"<?xml version=""1.0"" encoding=""UTF-8""?><Data Type=""Query"" Version=""2.1""><Query Product=""Test""></Query></Data>";
I couldn’t find a good way to switch this encoding on my end in order to be able to save to the database, since SQL Server expects UTF-16 data. I tried to switch the encoding to UTF-16(Unicode) using the below cod but when I try to save to the database get an “illegal xml character” error. Moreover the data in alteredText looks like garbage data.
MemoryStream memStream = new MemoryStream();
StreamWriter swUtf16 = new StreamWriter(memStream, Encoding.UTF8, Test_PSQ1.Length);
swUtf16.WriteLine(Test_PSQ1);
int length = Convert.ToInt32(memStream.Length);
byte[] buffer = new byte[length];
memStream.Write(buffer, 0, length);
memStream.Seek(0, SeekOrigin.Begin);
String alteredText = System.Text.Encoding.Unicode.GetString(memStream.GetBuffer());
I did some investigation and found that, If I can remove “<?xml version=""1.0"" encoding=""UTF-8""?>” part from the XML string, the save would work fine. I am trying to do some XML parsing so that I can strip out this tag, but am unable to do so.
MemoryStream memStream = new MemoryStream();
StreamWriter swUtf16 = new StreamWriter(memStream, Encoding.Unicode, Test_PSQ1.Length);
swUtf16.WriteLine(Test_PSQ1);
int length = Convert.ToInt32(memStream.Length);
byte[] buffer = new byte[length];
memStream.Write(buffer, 0, length);
memStream.Seek(0, SeekOrigin.Begin);
XmlReader reader = new XmlTextReader(memStream);
test = reader.ReadElementString();
test = reader.Value;
if (reader.NodeType == XmlNodeType.Element)
{
test = reader.Value; ;
}
reader.Read();
if (reader.NodeType == XmlNodeType.Element)
{
test = reader.Value; ;
}
Any suggestions about the code, pointing to resources would be highly appreciated.
|
|
|
|
|
This should work:
XmlReader reader = new XmlTextReader(memStream);
reader.MoveToContent();
string NewXmlText = reader.ReadOuterXml();
topcoderjax - Remember, Google is your friend.
|
|
|
|
|
Hello everyone,
I am trying to place an image into a button. The Image originally is placed in the DataGridView and using the following code I am placing it in the desited button.
btnItemImage0.Image = (Image)dataGridView1.Rows[ItemRowIndex0 - 1].Cells["item_image"].FormattedValue;<br />
The images have different sizes (They are mainly bigger than the button size!). Can someone tell me how I can resize the image to make sure the entire image is visiable in the button?
Thank you very much and have a great day.
Khoramdin
|
|
|
|
|
You can create instance of Bitmap class and specify sizes
|
|
|
|
|
I am having problems saving changes made in a datagrid to a database. I am using SqlDataAdapters with a generic "Select * FROM tbl" to load a dataset. I then assign the one table at a time to the control listing the contents of the table (all the tables are lookups for the application). Everything works great up to the point that the data is not being saved back to the server. It looks like it works. The problem is that it does not appear that the values in the datagrid are loaded back to the datatable. Here is how I am loading the grid (I am assuming that I am missing something basic here):
<br />
SqlDataAdapter adp = new SqlDataAdapter("SELECT * FROM [" + sTmp + "];", dbConn);<br />
dbTableAdp.Add(adp);
adp.FillSchema(dbSet, SchemaType.Source, sTmp);<br />
adp.Fill(dbSet, sTmp);<br />
adp.AcceptChangesDuringUpdate = true;<br />
adp.ContinueUpdateOnError = true;<br />
<br />
DataTable dt = dbSet.Tables["tablename"];<br />
if (dt != null)<br />
{<br />
dgLookups.DataSource = dt;<br />
}<br />
<br />
if (row.IsCurrentRowDirty == true)<br />
{<br />
DataRow dr = dbSet.Tables["tablename"].Rows[e.RowIndex];<br />
if (MessageBox.Show("Do you wish to commit changes to the database (Yes/No)?",<br />
"Values Changed", MessageBoxButtons.YesNo, MessageBoxIcon.Question) ==<br />
DialogResult.Yes)<br />
{<br />
DataGridViewRow dgr = dgLookups.Rows[e.RowIndex];<br />
dgr.SetValues();<br />
dr.AcceptChanges();<br />
}<br />
else
{<br />
dr.RejectChanges();<br />
}<br />
}<br />
I am fairly new to using the DataGrid. I have tended toward using ListView since I normally don't allow direct access to the database. Any assistance will be a great help.
Leo T. Smith
Senior Programmer
|
|
|
|
|
Try this.
Smile: A curve that can set a lot of things straight!
(\ /)
(O.o)
(><)
|
|
|
|
|
Hi,
I developed a Transactional Server that receives at leat a transaction per second.
The application has a pool of threads that resolve each incoming transaction. This threads use reflection for dynamic execution of generated source code. If the execution returns 0 the program send an approval. If the executions returns 0, the application creates a thread to notify the adminitrators of the systems.
The thread join was resolve creating a CollectorThread that receives a reference of every thread and it wait to them to die.
The application is working well, except for one issue. When each thread uses the Reflection to create the Dinamic execution the memory grows about 80kb, after that I disposed the object and wait for the garbage collector to clean that memory.
When the app is running in Windows XP everything seems to work just fine. But, when the app is running in Windows 2003 ( Dual Core 2 ) the memory grows and grows and it's never collect by the GC.
I also try the AppConfig GCServer option, but it didn't fixed the issue.
Somebody knows something about a bug with the Framework 2.0 in Windows 2003? Is there something else that reflection? (Besides programming my own parser)
Thanks.
rmga
|
|
|
|
|
Hi,
gc runs automatically when it needs to run (i.e. when free memory is needed and
not available); if your Win2003 machine has more memory,
or your process is allowed a larger working set, gc will not run (or only much
later).
So my guess would be there is no problem at all. You could try to (dis)prove this
by reducing the working conditions e.g. adding other, unrelated processes, and/or
changing system parameters reducing the default or max working set (which may
be larger on any Windows server).
BTW I do not recommend calling gc explicitly !
|
|
|
|
|
Hi,
Thanks for the advice. We hope that there are no problems with the code and it is just a configuration issue. Just a doubt, where can I change the max working set?
Thanks,
rmga
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
