2. Histogram, Sorting and Saving:

This feature permits a user to sort and save as text, frequency histograms of all colors for the entire image or selected regions, by 10 different axes including: frequency, color index, hue, saturation, brightness, red, green, blue, mean RGB, and mean HSV. The user can visualize the image in several different modes and determine which of the 10 color axes "control" the organization. The user can use that axis to control the fuzzy searching and growing as described below.

3. Palette Animation and Classification:

A user can highlight and animate colors by blinking them individually or by groups. If colors are grouped, then the relative abundance of the group is calculated and displayed. Palette animation allows full visualization of various patterns in an image. It also serves as a means of visualizing various image classification schemes.

4. Palette-based Image Processing:

The user can classify, process and group pixels by color, brightness, saturation, mean RGB, mean HSV, or frequency.

5. Spatial-based Image Processing:

Pixels can be classified and processed as a function of their spatial proximity so that neighboring pixels can be combined to create a new color group.

6. Heads-up Display:

FullPixelSearch supports a "heads-up" display whereby data pertaining to a region or a point in the image is displayed in a user-defined color to the right of the cursor as it ranges over the image. The user selects the data displays, which currently include X, Y coordinates, color index value, a user-entered text name, and RGB values.

7. Pixel Annotation:

This function enables users to annotate the image with descriptive text that is "tagged" to the color index value. As the heads-up display ranges over the image, the text names appear and are updated over each pixel. These names are stored in the image resource and can be exported from or imported into an image. Using pixel annotation in conjunction with the heads-up display is useful both for live presentations and storing image content information.

8. Shannon Diversity Index:

The Shannon Diversity Index is calculated for every histograms FullPixelSearch generates. Diversity (H'), is calculated as follows: H' = - … (pi ln pi), where pi is the proportion of color (i) in the current histogram and ln is the natural log of pi. The diversity index generates a single value usually between 0 and 5, for each histogram. The index increases as the number and relative abundance of colors increases. Thus, H' is a measure of diversity or "information" within a selected image or portion of an image. FullPixelSearch can compute a diversity image and render gray scale map show pockets of information.

2.2 Pixel Searching and Matching Features

1. Neighbor Search:

This function measures the neighborhood or bounding region, of any single color, up to a radius of 6 pixels from each target pixel. FullPixelSearch stores each of these histograms separately, thus allowing comparison.

2. Pattern Recognition & Pixel Matching:

FullPixelSearch achieves pattern recognition through pixel matching and fuzzy logic. The user identifies pixels to be searched for by copying a region to a template or by drawing a pattern directly into the template. The template is an 8 pixel by 8 pixel box where the "model" is constructed. Pixels within the template can be designated as "wildcards", meaning that any color will satisfy that location of the pattern, or by their actual value so they are excluded from the fuzzy search. Templates can be named and saved and thus reloaded to search in other images.

3. Search with Rotation:

FullPixelSearch supports searching for pixels in any combination of the 8 major directional axes.

4. Diversity Search:

Diversity Search lets the user search for areas of the image that have the greatest representation of colors from the image. This function is especially useful in earth image analysis and remote sensing where the ability to quickly identify areas of homogeneity or heterogeneity within an satellite image is of paramount importance.

2.3 Exact Matching Features

1. Exact Match:

Exact Match will search the image for exact matches of the current model in the template. This means that a positive hit for an exact match requires that an exact duplicate of the model be found somewhere in the image.

2. Exact R Match:

Exact Replacement Match is an Exact Match, like above, that allows replacements of colors. This means that any combination of the colors selected in the model, will satisfy an Exact R Match. For example, if a model contains 3 pixels of 3 colors in a row, any combination whatsoever, of these 3 colors will satisfy the Exact R Match.

3. Exact NR Match:

Exact No Replacement Match does not accept replacement. Rather, it preserves the relative abundance of pixels in the model . For example, if a template contains 3 colors each in equal proportions, only regions where these proportions are preserved will satisfy the search.

4. Exact Color Match:

Exact Color Match will find all the pixels selected in the model everywhere in the image.

2.4 Fuzzy Matching Features

1. Fuzzy Match:

The Fuzzy Match search extends the Exact Match search by adding colors that are either positive or negative of the colors selected in the model (e.g., more red or less red, brighter or darker, more saturated or less saturated, etc.). For example, if a precise grouping of pixels holds significance to the user, the Fuzzy Match search finds additional matches (inclusive of the exact matches) by a fuzzy amount that has been user-defined.

2. Fuzzy R Match:

Like the Exact R Match described above, Fuzzy Replacement Match will find any combination of the colors selected in the model but also include the fuzzy colors as well. If a model contains 3 different colors the Fuzzy R Match will find the exact matches and the fuzzy matches.

3. Fuzzy NR Match:

Fuzzy No Replacement Match does not accept replacement. Rather, it preserves the relative abundance of pixels selected in the model plus the valid fuzzy set. For example, if a template contains 3 colors each in equal proportions, only regions where these proportions plus their fuzzy set, is preserved will satisfy the search.

4. Fuzzy Color Match:

Fuzzy Color Match will find all the pixels selected in the model plus the fuzzy valid set, everywhere in the image.

2.5 Pixel Growing Features

1. Exact Grow:

Exact Grow results in the expansion of any current selection. The pixels which a user adds during exact grow are a function of the colors selected in the template. Exact growth only will add new pixels from current template providing that they share a border (a border is defined as a shared side not a corner) with an identical color in the current selection. Exact growing occurs along pathways of exact matches to colors found at the perimeter of a current selection. Growing occurs in two modes, bounded and unbounded. Bounded growing will add pixels to the current selection out to a user-defined distance specified in number of pixels. Unbounded growing will continue outward infinitely until no more growth is possible or the edge of the image is reached. Exact growing allows the exploration of contiguity of colors and objects in an image. Because exact growing occurs only when similar colors border on each other, the amount of growing that results is a direct measure of contiguity or connectivity among colors.

2. Exact R Grow:

Exact Random Grow results in the expansion of any current selection because it is not limited to shared borders of identical colors. Like Exact Grow, it occurs in two modes, bounded and unbounded. The pixels added during Random Exact Grow are a function of whatever colors are selected in the template. Random Exact Grow will add new pixels from this list in the template anywhere that any of them share a border with any pixel in the current selection. Therefore, random exact growing occurs outward from the current selection to include any bordering pixels specified in the template.

3. Fuzzy Grow:

Fuzzy Grow results in the expansion of any current selection and is a function of the exact colors selected in the template and the additional colors selected by the fuzzy options. Fuzzy growth only will add pixels from the template or from the fuzzy list provided that they share a border with an identical color in the current selection. Fuzzy growing thus occurs along pathways of exact matches to colors found at the edge of a current selection. Fuzzy growing allows the exploration of contiguity of colors and objects in an image thereby working from a longer list of potential colors. It usually results in more growth, therefore, than exact or random exact growing. Fuzzy growing is spatially constrained, however, in the direction of pixels that share a boundary with other identically colored pixels.

4. Fuzzy Random Grow:

Fuzzy Random Grow results in the expansion of any current selection and is a function of the exact colors from the template and the additional colors made relevant by the fuzzy options. Random fuzzy growth will add new pixels specified in the template or from the fuzzy list anywhere along the border of the current selection. Random fuzzy growing thus occurs outward from the current selection to include any bordering pixels specified in the template and fuzzy options. Random fuzzy growing allows the exploration of contiguity of colors and objects in an image and because it works off a longer list of potential colors and is not spatially constrained, random fuzzy grow results in more growth than any other type of growing.

---

FPS

Quick Ref

Basic Oper

Resume

Next