I don't have sample code, as a fair amount of code would be required to do this. I can point you to the API functions you should be calling though, and in what order.

1. First you need a DocumentInfo instance that represents the document you'll be working with. If you are using Workflow, then an EntryInfo reference representing the entry should be passed to the script. Cast this to a DocumentInfo.
2. You need to iterate through the pages so you can retrieve the text and image data. Call DocumentInfo.GetPageInfos, which will return a PageInfoReader instance. PageInfoReader implements IEnumerable and you should use a loop to iterate through the pages by retrieving a PageInfo instance for each page. You will be processing one page at a time.
3. Call PageInfo.ReadTextPagePart and read the text part of the page out. Store the text into a local MemoryStream instance you allocate.
4. Instantiate a WordsReader instance from the MemoryStream that contains the page text and call WordsReader.Read in a loop until you find a PageTextWord value that represents the key word you're looking for, such as "TEST".
5. Reset the MemoryStream.Position property for the text page to 0 to reset the stream and create a new instance of WordsReader.
6. Call PageInfo.ReadLocationsPagePart to retrieve a WordLocationsReader instance. The locations data for the page contains data about where the on the image each word is located. This will be used to find a corresponding rectangle and its coordinates on the image that is the bounding box for the key word.
7. Instantiate a TextLinker instance, passing in the WordsReader instance from step 5 and the WordsLocationReader from step 6 to the constructor.
8. Call TextLinker.GetRectanglesInRange, passing in as the start and end range the starting and ending position for the key word "TEST" that is available as properties of the PageTextWord value you retrieved in step 4. You now have a rectangle listing the coordinates of the word "TEST" on the image portion of the page.
9. Call PageInfo.ReadPagePart, specifying PagePart.Image to retrieve a Stream instance representing the image portion of the page.
10. Reference the LaserficheImaging.dll assembly. Instantiate an instance of Laserfiche.Imaging.Core.LfWriteableBitmap, passing in the Stream instance from step 9 to the constructor.
11. Query the properties of LfWriteableBitmap to determine the image's width and height.
12. Now you can split the source image into several smaller images. For each smaller image you want to create, instantiate a new instance of LfWriteableBitmap, specifying the desired width, height, DPI and pixel format. You should copy the DPI and pixel format values from the properties of the source (original) LfWriteableBitmap.
13. Call LfWriteableBitmap.WritePixels on each of the destination (new) LfWriteableBitmap instances, specifying the coordinates in the source LfWriteableBitmap to copy over.
14. Instantiate an instance of Laserfiche.Imaging.LfiBitmapEncoder, specifying the desired image format. For each new bitmap, set the element of LfiBitmapEncoder.Frames to reference the LfWriteableBitmap instance representing the new bitmap and call LfiBitmapEncoder.Save to save a copy of the image to disk. You will later delete this temporary file.
15. Call DocumentInfo.InsertPage to create a new page and store the returned PageInfo reference representing the new page. You might want to do this in a new document so as not to disturb the source document.
16. Call PageInfo.WritePagePart specifying PagePart.Image to retrieve a Stream instance that you will use to upload the image data. The size argument to WritePagePart should be the length of the image file that was saved to disk in step 14.
17. In a loop call Stream.Write to upload the data from the image file you created in step 14 to the page you created in step 15. When you're done, delete the temporary file.
18. You're done. Note that steps 12-17 will have to be repeated for each sub-image you create from the original image.