<?xml version="1.0" encoding="utf-8"?>
<feed xml:lang="en" xmlns="http://www.w3.org/2005/Atom"><title>Recent changes to feature-requests</title><link href="https://sourceforge.net/p/fourier-ipal/feature-requests/" rel="alternate"/><link href="https://sourceforge.net/p/fourier-ipal/feature-requests/feed.atom" rel="self"/><id>https://sourceforge.net/p/fourier-ipal/feature-requests/</id><updated>2009-07-08T23:31:14Z</updated><subtitle>Recent changes to feature-requests</subtitle><entry><title>Add math operations + image stats</title><link href="https://sourceforge.net/p/fourier-ipal/feature-requests/3/" rel="alternate"/><published>2009-07-08T23:31:14Z</published><updated>2009-07-08T23:31:14Z</updated><author><name>Anonymous</name><uri>https://sourceforge.net/u/userid-None/</uri></author><id>https://sourceforge.net1bcfb9dd3fe16f0a7baf2506ca2543dfa42c3d3a</id><summary type="html">&lt;div class="markdown_content"&gt;&lt;p&gt;Please add the following basic math operations&lt;br /&gt;
- image_MIN( image * i, const int min_value)  ==&amp;gt; Set any pixel below MIN_VALUE to MIN_VALUE&lt;br /&gt;
- image_MAX( image * i,  const int MAX_VALUE) ===&amp;gt; Set any pixel above MAX_VALUE to MAX_VALUE&lt;br /&gt;
- image_MINMAX ( image * i, const in MIN_VALUE, const int MAX_VALUE )  ==&amp;gt; combination of image_MIN and image_MAX functions&lt;/p&gt;
&lt;p&gt;- image_SUB_TO_ZERO ( image *i ) ===&amp;gt; find minimum value in image and then subtract it from entire image to set the minimum value to 0.  This is useful when one knows beforehand that useful image data is in the lower part of the histogram&lt;/p&gt;
&lt;p&gt;- image_ADD_TO_MAX ( image *i) ===&amp;gt; Same as image_SUB_TO_ZERO() but add a number to each pixel so that the maximum pixel value is MAX_GRAY&lt;/p&gt;
&lt;p&gt;- image_GET_STATISTICS (image *i,  /*output */ statistics *s) ===&amp;gt; get the follwoing information about the image in a structure&lt;br /&gt;
- minimum value, maximum value, difference between min and max valuees&lt;br /&gt;
- minimum possible value, maximum possible value, difference between minimum possible value and maximum possible values&lt;br /&gt;
- average&lt;br /&gt;
- mode&lt;br /&gt;
- standard devaition&lt;br /&gt;
- histogram&lt;br /&gt;
- cumulative histogram&lt;br /&gt;
- number of unique colors&lt;br /&gt;
- number of unique grey values (intensity values if a color image)&lt;/p&gt;
&lt;p&gt;Examples of usage&lt;/p&gt;
&lt;p&gt;1) Emphasis on lower end of histogram&lt;br /&gt;
- subtract integer X from imaage pixels so that image has minimum value of 0&lt;br /&gt;
- apply LUT (exponential or other) to recolor the image data so that data near the lower end (near 0) of the image is emphasized&lt;/p&gt;
&lt;p&gt;2) Trim outer bounds of image data before normalization.  This simplifies the image for further processing  since, for many images, the useful data is in the middle part of the histogram.  Reducing dynamic range of the histogram (without affecting interior values) is useful before further processing (e.g., equlization, normalization, thresholding).&lt;br /&gt;
if (standard_deviation &amp;gt;  (25% of image dynamic range) then&lt;br /&gt;
{&lt;br /&gt;
- Set values lower than AVERAGE - 1.5 standard devaiations to (AVERAGE - 1.5 standard deviations)&lt;br /&gt;
- Set values above AVERAGE + 1.5 standard devaitions to (AVERAGE + 1.5 standard deviations)&lt;br /&gt;
}&lt;br /&gt;
- Normalize image&lt;/p&gt;
&lt;p&gt;3) Set the pixels in the bottom 25% of the histogram to black.  This is equivalent to thresholding the image so that 25% of the pixels are black and then overlaying the thresholded image on top of the original image.  This helps define lines and edges before calling an edge method (Canny) and has applications in detecting cell boundaries or OCR pre-processing.&lt;/p&gt;
&lt;p&gt;The direction here is that higher level image processing methods help greatly but lead to difficulty in determing the right input parameters (option overload problem).  Processing the image with simpler operations first before applying higher level methods will greatly simplify the number of choices one needs to make in selecting parameters.  It is often easier to determine that useful data for a given image has intensity values from A to B and use that range as a starting point for further processing than it is to fine tune the input parameters to a higher level method to achive the end user's desired result.&lt;/p&gt;
&lt;p&gt;p.s. Thank you for the thresholding method example code.  It leads one to the supposition that the various thresholding algorithms can be combined in polynomial fashion to better threshold an image.&lt;/p&gt;
&lt;p&gt;This leads one to propose that one could use neural net, expert system, etc to automatically generate a better thresholding algorithm that takes as input&lt;br /&gt;
- original image&lt;br /&gt;
- output of niblack&lt;br /&gt;
- output of thresholding algorithm X1&lt;br /&gt;
- output of thresholding algorithm X2&lt;br /&gt;
- output of thresholding algorithm X3&lt;br /&gt;
- output of thresholding algorithm X4&lt;/p&gt;
&lt;p&gt;This is similar to how OCR engines have many lower level OCR engines that VOTE for the output of each character.&lt;/p&gt;&lt;/div&gt;</summary></entry><entry><title>Extract objects found in binary image</title><link href="https://sourceforge.net/p/fourier-ipal/feature-requests/2/" rel="alternate"/><published>2009-05-14T16:40:11Z</published><updated>2009-05-14T16:40:11Z</updated><author><name>Anonymous</name><uri>https://sourceforge.net/u/userid-None/</uri></author><id>https://sourceforge.net6f46837bd261df08cedca1714591f17c826f5f1c</id><summary type="html">&lt;div class="markdown_content"&gt;&lt;p&gt;New feature suggestion: &lt;/p&gt;
&lt;p&gt;Extract objects found in binary image.  Each object is a connected set of black pixels not connected to another object.&lt;br /&gt;
&lt;/p&gt;
&lt;p&gt;An object contained within another object but not touching that object would be a seperate object.&lt;/p&gt;
&lt;p&gt;Output format would be:&lt;br /&gt;
- A different image file for each object found (png, tiff, bmp, but not lossy jpeg)&lt;br /&gt;
- Pixels in the image file for an object are black for just that object but not any other pixels (i.e., the output contains only the object found and not nearby black pixels connected to another object).&lt;br /&gt;
- Output file name would contain the location of the object's upper left hand pixel and the original dimensions of the input image.  For example, a symbol similar to a scanned 'P' 25x15 pixels with upper left hand pixel at 100,200 would have filename 00100_00200_00025_00015.png.  This could be extended to include the count of the black pixels, the center of gravity or other symbol measures.  Encoding them in the filename yields to easier processing in other applications since one can send the list of filenames to a text file and import into a spreadsheet or a scientific application without any extra custom coding.&lt;/p&gt;
&lt;p&gt;To reconstruct the original image:&lt;/p&gt;
&lt;p&gt;output_created = false&lt;br /&gt;
For each file in the directory containing the objects &lt;br /&gt;
{&lt;br /&gt;
1. Get file name&lt;br /&gt;
2. Find X,Y of upper left hand pixel from the file name&lt;br /&gt;
3. If output_created == false then&lt;br /&gt;
{&lt;br /&gt;
get the output image dimensions from the file name&lt;br /&gt;
create a blank image of that size (e.g., all white pixels)&lt;br /&gt;
}&lt;br /&gt;
3. Overlay the black pixels only on the output image at position (x + row in the object image, y + column in the object image) &lt;br /&gt;
}&lt;br /&gt;
Write output image&lt;/p&gt;
&lt;p&gt;This has many uses like blob counting, object dection after filtering out overly large/small objects, shape analysis, etc.&lt;/p&gt;&lt;/div&gt;</summary></entry><entry><title>EM_INVALID</title><link href="https://sourceforge.net/p/fourier-ipal/feature-requests/1/" rel="alternate"/><published>2009-05-01T18:30:01Z</published><updated>2009-05-01T18:30:01Z</updated><author><name>Anonymous</name><uri>https://sourceforge.net/u/userid-None/</uri></author><id>https://sourceforge.net1b399133a8262dd6fe22b996f9a48d36c33e73fd</id><summary type="html">&lt;div class="markdown_content"&gt;&lt;p&gt;Can the EM_INVALID found in the enumerated types be changed to something else.&lt;/p&gt;
&lt;p&gt;It causes a conflict with Microsoft Visual Studio 2005 float.h include file.  A google search for "EM_INVALID" shows it is in the boost C++ libraries as well.&lt;/p&gt;
&lt;p&gt;Can this define be changed to not conflict with MS Visual Studio's float.h include file?&lt;/p&gt;
&lt;p&gt;#define EM_INVALID              _EM_INVALID&lt;/p&gt;&lt;/div&gt;</summary></entry></feed>