|
|
Page Detection and Margins
| Page Detection and Margins
| |
|
|
| Determining Image and Margin Size
|
| The Margin settings provide a useful and flexible way to check your images for irregularities.
| |
| Margins can be set relative to the edges of the pages. Set the left-right and top-bottom margins dynamically for "Set" or "Check". If one margin is "Set" to a certain measurement, the other side will automatically change to "Check". After detecting the subimage and positioning it to the to "Set" margins, the "Check" margins will then be examined. The question is: is there any margin "Check" criteria violated? The subimage should not stick out of the margins, otherwise an exception will be generated.
| |
| An average publication, book or form will have similar overall image size. The goal is to set up a margin criteria, which is just large enough to fit into the margins, but tight enough to catch irregularities around the edges.
| |
| Consider the example in Figure 5-1, which is a scanned page from a book. All pages in this book are similar, with a predictable image size of approximately 4.4 x 7.6" on a 5.76 x 9.07" canvas size. This gives a resulting horizontal margin (Left+Right)=1.36" and resulting Vertical margin (Top+Bottom)=1.47".
| |
| A convenient setup for a Recto page will be set up as shown in Figure 5-2. Pay attention to the "Set" criteria at the Top and Right margins. The Left and Bottom margins are automatically set to "check".
| |
| After processing to the above parameters, the image did not pass and ended up in the Exception List with the message "Vertical Margin Error". Figure 5-3 shows a portion of the Investigate screen and the resulting view after processing. The positioned image is shown in Figure 5-3 and the margins are drawn with dashed lines. Detail "A" shows a large, irregular speckle, whereas Detail "B" shows an average speckle, which will be safely filtered out by the page detection parameters.
| |
| Figure 5-3. Positioned Image with Vertical Margin Violation.
| |
| Examining the image through the Investigation Tool and using the "Focused Zoom" feature, you can see the effects of filtering on Details "A" and "B":
| |
| In Figure 5-3, Detail "A" clearly shows that the large speckle is in the size range of characters. As shown in Figure 5-4, PPP did not remove it as a throwaway object, but considered it a part of the image. This is a safe approach, because if this size was included in the page detection algorithm, some character and image damage would occur. Now it is safe to edit the image using PixEdit, save it, and re-process it using the same parameters.
| |
| Figure 5-4. Detail "A" with Large Speckle
| |
| As a comparison, Detail "B" in Figure 5-3 shows a small speckle on the lower left of the image. This is correctly thrown out and not considered as part of the image as shown in Figure 5-5.
| |
| Figure 5-5. Detail "B" with Small Speckle
| |
| The above examples show that it is very useful to pay close attention to the margin size. If tight margins are used, so tight that the detected subimages fit exactly into them, a lot of exceptions will be generated, thus a lot of images to review and approve. In contrast, one can set up loose margins. In an extreme case, set all margins to zero and position the subimage to the center of the output canvas. In this case, a lot of free space is between the margins, so the subimage can be much larger than desired, and the image processor will not generate an exception, unless the subimage is larger than the output canvas.
| |
| Needless to say, none of these two extremes are practical. It is not very convenient if all of your images are found erroneous. If bad images pass the quality check, that is definitely wrong. A skilled operator will always find the right combination of margins, where only the good images will be passed to the Good Files List.
| |
| If the size of the output canvas is the same or less than the size of the input image, there is no need to set up any other parameters, PPP will do the quality control. But if the size of the output canvas is larger than the input image, the Maximal Subimage size parameter will be have to be used. By specifying maximum detected subimage size, the same effect can be applied to the image as it was a margin criteria.
| |
© 1998-2002 ELAN GMK
|
