EP2569932A1

EP2569932A1 - Method and apparatus for color enhancement

Info

Publication number: EP2569932A1
Application number: EP11725200A
Authority: EP
Inventors: Erno Langendijk; Andreas Stefan Hotz; Karel Hinnen
Original assignee: TP Vision Holding BV
Current assignee: TP Vision Holding BV
Priority date: 2010-05-10
Filing date: 2011-05-04
Publication date: 2013-03-20
Also published as: WO2011141853A1

Abstract

An apparatus for performing a color enhancement of an image comprises a segmenter (103) which generates image segments that specifically may be relatively small. An analyzer (113) identifies a neighbor segment for a first segment and a color enhancer (105) applies a color enhancement algorithm to the first segment. An adjuster (107) is arranged to adjust a characteristic of the color enhancement algorithm for the first segment in response to a relative geometric property of a resulting group of color points in a color space and a neighbor group of color points in the color space. The resulting group of color points comprises a color point for at least some color enhanced pixels of the first segment. The neighbor group of color points comprises a color point for at least some pixels of the at least one neighbor segment. The segmentation based color enhancement considering inter-segment color properties may provide improved image quality.

Description

Method and apparatus for color enhancement

FIELD OF THE INVENTION

The invention relates to color enhancement and in particular, but not exclusively, to color enhancement for televisions. BACKGROUND OF THE INVENTION

Color enhancement of images, and in particular of digital images, has become widespread and is used in many display devices, for photo manipulation, special effects etc. For example in today's digital color televisions, the colors are typically enhanced in order to make more appealing images.

Various methods are known for color enhancement of images. Examples may for example be found in Jan Morovic, M. Ronnier Luo , "Evaluating Gamut Mapping Algorithms for Universal Applicability", COLOR research and application, vol.26, pp. 85- 102, 2001.

Typically for such color enhancement algorithms, the color saturation is increased by a constant factor for each input pixel such that a general saturation increase results from the saturation increase of the individual pixels. One problem associated with many conventional saturation enhancement algorithms is that if the saturation increase is set too high, a certain amount of clipping may be introduced. Although this may be acceptable in many scenarios, it may also be disadvantageous for many images. In particular, it may result in some pixels that have different colors in the input image will be saturation enhanced to the same color points, and thus will be rendered as the same output color. Such clipping is particularly disadvantageous as it tends to result in very noticeable artifacts, such as contouring and a loss in detail. Indeed, typically areas with relatively small color variations may be color enhanced to identical colors and they will therefore appear as a single homogenous image area rather than a detailed and possibly textured area. Such loss of details tends to be relatively clearly noticeable.

Hence, an improved color enhancement would be advantageous and in particular an approach allowing increased flexibility, facilitated implementation or operation, reduced artifacts or perception thereof, improved image quality and/or improved performance would be advantageous.

SUMMARY OF THE INVENTION

Accordingly, the Invention seeks to preferably mitigate, alleviate or eliminate one or more of the above mentioned disadvantages singly or in any combination.

According to an aspect of the invention there is provided an apparatus for performing a color enhancement of an image, the apparatus comprising: a segmenter for generating image segments for the image; an analyzer for identifying at least one neighbor segment of a first segment in accordance with a neighborhood criterion; a color enhancer for applying a color enhancement algorithm to the first segment; a processor for determining a resulting group of color points in a color space comprising color points for at least some color enhanced pixels of the first segment; a processor for determining a neighbor group of color points comprising color points for at least some pixels of the at least one neighbor segment; andan adjuster for adjusting a characteristic of the color enhancement algorithm for the first segment in response to a geometric property of a distribution of color points of the resulting group of color points relative to a geometric property of a distribution of color points of the neighbor group of color points..

The invention may provide improved color enhancement. In particular, it may alleviate visible artifacts introduced by color enhancement and in particular may mitigate loss of image detail or contouring that often result from clipping when performing color enhancement.

The inventors have realized that color enhancement may advantageously be based on a segmentation approach that does not only consider individual segments or pixels but which also consider relative color space characteristics for neighboring segments. In particular, the approach may allow a mitigation of noticeable artifacts introduced by e.g. clipping.

For example, the approach may allow the color enhancement to be adapted to the specific characteristics of each segment such that a more localized optimization of the image enhancement can be achieved without risking the introduction of unfortunate intersegment artifacts. As a specific example, the approach may allow different segments to be enhanced within a limited gamut without risking the segments becoming too similar due to clipping. In particular, non overlapping segments may remain non-overlapping thereby retaining visual separation from the original image. The color enhancement algorithm may be monochromatic but does not need to be. Thus, in some embodiments the color enhancement algorithm may modify luminance as well as chromaticity. In other embodiments only chromaticity may be modified. The color space may be any suitable color space such as an RGB color space, or a luminance/ chromaticity color space such as the standardized xyY color space. The color space may be any N dimensional color space where N=l , 2, 3... The color space may be a standardized or non-standardized color space. Specifically, the color space may be a two dimensional chromaticity only color space (corresponding to a chromaticity plane) or may be a three dimensional color space such as a chromaticity and luminance color space. As another example, the color space may be determined by primary colors such as an RGB color space. The color space may for example be a mixed color space where the different axes expanding color space may represent contributions from two or more of the hue, chroma and luminance parameters.

The image segments may be contiguous but need not be. The image segments may typically relate to relatively small areas or pixel groups with very close image properties. The image need not correspond to image objects but may often be substantially smaller. In particular, the image segmentation may be what is typically referred to as an over- segmentation, i.e. it may generate image segments sufficiently small to divide image objects into multiple segments. The color enhanced pixels may be pixels enhanced by the application of the color enhancement algorithm.

The characteristic may be any value, property or control function/operation that may affect the color enhancement performed. In particular, it may be one or more parameters.

In accordance with an optional feature of the invention, the color points of the neighbor group of color points correspond to color enhanced pixels.

This may provide improved performance in many embodiments. In particular, it may allow that the desired color relationships are maintained in the enhanced image by directly considering color space characteristics for segments in the color enhanced images.

The color enhanced pixels may specifically be pixels resulting from

application of the color enhancement algorithm to the neighbor segment, possibly using a different characteristic than for the first segment. For example, the color enhancement algorithm may first be applied independently to the first and neighbor segments. Based on the resulting groups of color points, the characteristics/parameters of the color image enhancement may be modified by the adjuster such that modified color enhancements can be generated with the desired relative color space characteristics.

In accordance with an optional feature of the invention, the adjuster is arranged to control the characteristic to restrict an overlap of color points between the first segment and the neighbor segment to be less than a threshold.

This may provide improved color enhancement and may in particular mitigate color enhancement artifacts such as loss of detail. For example, it may ensure that color variations are maintained between segments thereby maintaining visual differentiation between these.

The threshold may be an absolute threshold or may be a relative threshold. For example, the threshold may require that the fraction of overlapping color points is less than a given fraction of the total number of color points in the first set and/or the second set. An overlapping color point may be one that is included in both the first and neighbor groups. As another example, the threshold may simply be a fixed number of pixels.

Specifically, the adjuster may be arranged to control the characteristic to avoid an overlap of color points between the first segment and the neighbor segment.

In accordance with an optional feature of the invention, the apparatus further comprises a processor for determining a first outline for at least the resulting group of color points and a second outline for the neighbor group of color points, and wherein the adjuster is arranged to control the characteristic to avoid an overlap between the first outline and the second outline.

The outline may for example be defined to be an overlap meeting a geometric requirement and for which a specific number of color points of the first and/or neighbor groups are included.

In accordance with an optional feature of the invention, the neighbor segment is closer to a gamut limit than the first segment and the color enhancer is arranged to apply a color enhancement to the neighbor segment to move this in the color space towards the gamut limit; and the adjuster is arranged to determine the characteristic such that the first segment is in the color space moved towards the gamut limit while restricting an overlap between the resulting group of color points and the neighbor group of color points below a threshold.

This may provide improved color enhancement and may in particular allow an improved enhancement of the image as a whole. In many scenarios it may allow an increased saturation of the enhanced image while reducing clipping and mitigating loss of detail.

In accordance with an optional feature of the invention, the apparatus is arranged to apply the color enhancement algorithm to the first segment with the characteristic having a first chromaticity modification value, and wherein the adjuster is arranged to determine a modified chromaticity modification value for the characteristic in response to a detection of a conflict between the neighbor group and the resulting group for the first chromaticity modification value, and wherein the color enhancer is arranged to redo the color enhancement by applying the color enhancement algorithm to the first segment using the modified chromaticity modification value.

This may provide improved enhancement and may provide an efficient implementation and/or operation. In many embodiments, the approach may result in a reduced computational burden. The chromaticity modification values may reflect or indicate the chromaticity impact of the application of the color enhancement algorithm to the first segment.

In accordance with an optional feature of the invention, the color enhancer is further arranged to apply the color enhancement algorithm to the neighbor segment to generate the neighbor group; and the adjuster is arranged to adjust a further characteristic of the color enhancement algorithm for the neighbor segment in response to the geometric property of the distribution of color points of the neighbor group of color points relative to the geometric property of the distribution of color points of the resulting group of color points.

This may provide improved enhancement and may in particular allow an improved optimization of the color enhancement as the color enhancement of both segments may be optimized depending on the segments themselves as well as the other segment. The approach may in particular in many embodiments provide a more natural looking image following the enhancement.

In accordance with an optional feature of the invention, the apparatus further comprises an outline processor for determining an outline for each of the image segments in the color space, an order processor for ordering the image segments in order of saturation, and a controller for sequentially applying the color enhancement algorithm to the image segments in order of decreasing saturation, and wherein the adjuster is arranged to adjust the characteristic in response to the outline of the first segment and the outline of the neighbor segment where the neighbor segment is a previously processed image segment.

This may provide a particularly advantageous color enhancement of the image as a whole.

In accordance with an optional feature of the invention, the neighborhood criterion comprises a requirement that the first segment and the neighbor segment belong to a same image object.

This may provide an improved image quality. In particular, the approach may allow an improved adaptation of segmentation based color enhancement that reflects particularly critical and noticeable artifacts.

The image segmenter may specifically be arranged to generate the image segments and combine these into image objects. Thus, each image object will typically comprise a plurality of image segments and the apparatus may be targeted to maintain inter- segment color relationships for such image segments thereby providing a perceptually more natural looking image segment.

In accordance with an optional feature of the invention, the apparatus further comprises a processor for determining an outline in the image for each of the image segments, and wherein the neighborhood criterion comprises a requirement that the first segment and the second segment have less than a threshold of intervening outlines.

The feature may allow a particularly advantageous identification of segments for which relative characteristics may be particularly noticeable.

The threshold may particularly be two, including the outlines of the first segment and the neighbor segment. The requirement may e.g. require that the first and neighbor segments are adjacent segments by requiring that there are no other intervening outlines between the segments than the outlines of the segments themselves.

In accordance with an optional feature of the invention, the neighborhood criterion comprises a requirement that a distance between an image point of the first segment and an image point of the segment is less than a threshold. This may provide an improved image quality. In particular, the approach may allow an improved adaptation of segmentation based color enhancement that reflects particularly critical and noticeable artifacts.

The distance may be determined as a relative distance, such as e.g. a distance relative to a size of the first and/or neighbor segments.

In accordance with an optional feature of the invention, the adjuster is further arranged to adjust the characteristic in response to relative characteristics for the color points of the resulting group.

This may provide particularly advantageous performance and may in particularly provide improved image quality. For example, a more natural and detailed perceived image may be achieved.

In accordance with an optional feature of the invention, the adjuster is further arranged to adjust the characteristic in response to a spatio-colorimetric profile of the first segment.

This may provide particularly advantageous performance. For example, the approach may allow important internal variations and/or characteristics of a segment to be at least partially maintained while still allowing a color enhancement. This may in particular provide an improved detail level of the enhanced image.

In accordance with an optional feature of the invention, the adjuster is further arranged to adjust the characteristic in response to a spatio-colorimetric profile of the neighbor segment.

This may provide particularly advantageous performance. For example, the approach may allow important internal variations and/or characteristics of a segment to at least partially affect the color enhancement of another segment. This may in particular provide an improved consistency of the enhanced image. The feature may e.g. be particularly advantageous for scenarios where the first segment and the neighbor segment are part of the same image object.

In accordance with an optional feature of the invention, the color space is a chromaticity plane.

This may provide particularly advantageous performance with low resource requirements. For example, computational resource demands may be substantially reduced. Particularly advantageous performance may in many cases be achieved by the approach when employing relatively small segments.

In accordance with an optional feature of the invention, the color enhancement algorithm is a saturation enhancement algorithm.

The invention may provide a particularly advantageous saturation enhancement algorithm.

According to an aspect of the invention there is provided a method of color enhancement of an image, the method comprising: generating image segments for the image;

identifying at least one neighbor segment of a first segment in accordance with a neighborhood criterion; applying a color enhancement algorithm to the first segment;

determining a resulting group of color points in a color space comprising color points for at least some color enhanced pixels of the first segment; determining a neighbor group of color points comprising color points for at least some pixels of the at least one neighbor segment; and

adjusting a characteristic of the color enhancement algorithm for the first segment in response to a geometric property of a distribution of color points of the resulting group of color points relative to a geometric property of a distribution of color points of the neighbor group of color points.

These and other aspects, features and advantages of the invention will be apparent from and elucidated with reference to the embodiment(s) described hereinafter.

BRIEF DESCRIPTION OF THE DRAWINGS

Embodiments of the invention will be described, by way of example only, with reference to the drawings, in which

FIG. 1 is an illustration of an example of elements of a color enhancement apparatus in accordance with some embodiments of the invention;

FIG. 2 is an illustration of an example of color space groups for two segments prior to color enhancement in accordance with some embodiments of the invention;

FIG. 3 is an illustration of an example of color space groups for two segments following color enhancement in accordance with some embodiments of the invention; and

FIG. 4 is an illustration of an example of color space groups for segments in accordance with some embodiments of the invention; and

FIG. 5 is an illustration of an example of color space variations for image segments processed in accordance with some embodiments of the invention. DETAILED DESCRIPTION OF SOME EMBODIMENTS OF THE INVENTION

The following description focuses on embodiments of the invention applicable to a color enhancement in the form of a saturation enhancement and in particular to a saturation enhancement in the chromaticity domain. However, it will be appreciated that the invention is not limited to this application but may be applied to many other color enhancements including for example color enhancement algorithms that also involve luminance changes.

FIG. 1 illustrates an example of a color enhancement apparatus in accordance with some embodiments of the invention. The color enhancement apparatus may specifically be part of a color television, such as an LCD television, and may particularly be arranged to provide saturation enhancement for digital images to be displayed on the LCD television.

It will be appreciated that although the following description will focus on an embodiment of an LCD color display, the approach may be used with many other types of display and indeed with many other devices or functionality for rendering or processing a visual image. For example, the approach may be used for image processing in digital cameras or may be applied in connection with printing e.g. as part of printer driver applications.

The color enhancement apparatus comprises an image input 101 which receives a digital image to be enhanced. The image is specifically a digital image of a video sequence to be presented by the LCD television and comprises one set of color values for each pixel of the television.

The image input 101 is coupled to a segmenter 103 which is arranged to generate image segments for the image. In the example, a fairly detailed segmentation is performed resulting in a large number of relatively small segments (e.g. typically 1000 segments per image). Typically the segmentation results in image objects comprising a possibly large number of segments although that it will be appreciated that the following approach may be applied even with a low number of segments, including scenarios were only two segments are generated. The segmentation may result in a segmentation wherein the majority (typically more than 80%) of the generated image segments contain less than 2000 pixels, but typically more than 25 or 50 pixels. Such an approach is often referred to as an over-segmentation.

The approach described in the following has been found to be particularly advantageous in many embodiments when combined with such over-segmentation into relatively small image segments. Segmentation is typically based on grouping spatially close pixels into groups of pixels that have the same visual characteristics in accordance with a suitable criterion. It will be appreciated that many different segmentation algorithms will be known to the skilled person and that any suitable algorithm may be used without detracting from the invention. Examples of suitable segmentation algorithms may e.g. be found in D. Comaniciu and P. Meer, "Robust analysis of feature spaces: Color image segmentation",

1997 IEEE Computer Society Conference on Computer Vision and Pattern

Recognition, Proceedings, vol. 1, pp. 750-755, 1997.

The segmenter 103 is coupled to an image enhancer 105 which receives the segmented image and which proceeds to apply color enhancements to the image segments. In particular, the image enhancer 105 applies an image enhancement algorithm individually to each segment where characteristics of the algorithm may vary for the different segments. In particular, the image enhancer 105 may apply a color saturation enhancement algorithm individually to each segment. This approach may allow the enhancement algorithm to adapt to the local characteristics for each segment and this may lead to improved quality color enhancement. For example, the degree of color saturation increase may be adapted to the specific color characteristics of each image segment.

The color enhancement apparatus further comprises an adjuster 107 which is arranged to adapt the color enhancement algorithm for each individual segment. In the following, the operation will be described with specific reference to the image enhancement for a first image segment, but it will be appreciated that the approach may be applied in parallel or sequentially to a plurality of segments and indeed to all segments of the image.

The adjuster 107 is arranged to adapt the operation of the image enhancer 105 for the first image based on color characteristics for the first segment as well as for a neighbor segment. The neighbor segment may specifically be an adjacent segment to the first segment in the image. The color characteristics are evaluated in a color space by considering relative geometric properties in color space for the two segments, and the image enhancement is controlled such that the desired characteristics are achieved in color space for the first segment relative to the neighbor segment. The color enhancement is controlled such that a geometric property of the group of color points that correspond to the first segment after the color enhancement has a desired relationship relative to a geometric property of group of color points corresponding to the neighbor segment. For example, the geometric properties and relation may include consideration of the spread, the localization relative to white, asymmetry etc for the color points in the color space. Thus, the adjuster 107 comprises a first processor 109 which determines a resulting group of color points that correspond to the color points that result from the application of the color enhancement algorithm to the first segment. For each color enhanced pixel, the corresponding color point may be included in the resulting group. The resulting group of color points may be based on the color points resulting from applying a nominal or initial color enhancement to the first segment, i.e. the resulting group of color points may correspond to color points prior to the adjustment of the color enhancement characteristic. For example, if the resulting group of color points, when applying a color enhancement with a nominal or previous parameter/characteristic, does not result in a desired color space geometric relationship with the neighbor group of color points, the parameter/characteristic may be modified based on this determination. In other examples, the resulting group of color points may be that resulting from the application of the color enhancement algorithm after the adjustment has been implemented, i.e. it may correspond to the color points resulting from the final adjusted color enhancement. For example, the parameter may be calculated to optimize a match to a desired resulting group characteristic calculated from the neighbor group.

Similarly, the adjuster 107 comprises a second processor 111 which determines a group of color points that correspond to the color points for pixels in the neighbor segment. In some embodiments, the neighbor group of color points may be an initial group of color points, e.g. corresponding to the color points for the pixels of the second segment prior to any color enhancement thereof. For example, the enhancement algorithm may be controlled so that it does not result in the color points of the enhanced first segment overlapping color points of the original neighbor segment.

However, typically, the neighbor group corresponds to the color points corresponding to the pixels of the neighbor segment after the application of the color enhancement.

In some embodiments, the first and second processor 109, 111 may determine the groups to comprise a color point for each of the pixels in the respective segment.

However, in other embodiments, one or both of the groups may comprise color points for only a subset of the pixels. Thus, the first and second processor 109,111 may be arranged to perform a subsampling of the color points for the segments. E.g. only every second pixel may be considered or the groups may be reduced to only comprise color points that are not too near the average color point for the segment. Thus, in the apparatus of FIG. 1, the color enhancement is not just a static color enhancement of individual segments or driven only by the enhancement of one segment depending on the characteristics of that segment. Rather, the enhancement depends on a neighbor segment and this dependency is based on a geometric evaluation in color space. In particular, color point groups for neighbor segments are used to adapt the color enhancement algorithm. As will be demonstrated in the following, such an approach may provide substantial advantages and tends to provide improved color enhancement. In particular, it has been found to substantially improve image quality (and in particular detail levels) for color saturation algorithms.

As a specific example, the adjuster may be arranged to control the color enhancement algorithm such that an overlap of color points between the first segment and the neighbor segment is less than a threshold. In some embodiments, the color enhancement algorithm may be controlled such that there is no overlap of color points between the first segment and the neighbor segment.

Such an example is illustrated in FIG. 2 and 3. FIG. 2 illustrates a first group

201 of color points corresponding to the first segment prior to application of any color enhancement. FIG. 2 further illustrates a neighbor group 203 of color points corresponding to the neighbor segment prior to application of any color enhancement. In the example, the two groups are for clarity illustrated as two dimensional groups in a two-dimensional

chromaticity plane but it will be appreciated that the description may equally apply to a grouping in a three dimensional color space (e.g. further comprising a luminance dimension) where the groupings will correspond to three dimensional groupings (typically three dimensional volumes rather than two dimensional chromaticity areas).

The two segments are neighbor segments in the image, i.e. they are geometric neighbors in the image in accordance with a suitable neighbor criterion. As such, they may possibly have very different groupings in the color space domain and e.g. may be fairly distant groups in the chromaticity plane. This may for example be the case when the two segments belong to different image objects and therefore have very different colors. In such a case, the color enhancement of the two segments may simply follow a nominal or default color enhancement algorithm as there will be no undesired conflicts in color space.

However, very often, two neighbor segments have very similar chromaticity and luminance values. For example, they may correspond to slightly different color shades of the same object. In this case the groupings in color space are also likely to be very close. FIG. 2 illustrates an example where the first group 201 and the neighbor group are also adjacent and close in color space. It should be noted that due to the segmentation they are unlikely to have a significant initial overlap and indeed will often be disjoint groups. Indeed, this may be an inherent consequence of a color space based segmentation wherein pixels are grouped together in segments based on the color characteristics of the pixels.

In such a case, the color enhancement may often result in clipping which may result in the clipping of the color space color points to the same values. E.g. a default color enhancement may result in both groups comprising color points that exceed the available gamut 205. Accordingly, these may be clipped to the same chromaticity values resulting in the differentiation between the different segments being reduced or removed completely. This may result in substantially reduced image quality and in particular a perceived loss of detail.

However, the color enhancer of FIG.1 may allow the color enhancement to be adapted for the individual segments in response to relative geometric characteristics in color space between neighboring segments.

In the example of FIG. 2, the color enhancement of the first segment may specifically be adjusted such that there is no overlap in color space between the resulting enhanced segments. The requirement of no overlap between the groups of color points in color space may for example result in the color enhanced groups of FIG. 3. First, a maximum saturation increase may be provided to the neighbor segment 203 to result in an enhanced chromaticity grouping 301 that is adjacent to the gamut 205. The color enhancement

(specifically the saturation enhancement) may then be adapted for the first segment such that no overlap occurs between the enhanced chromaticity grouping 301 and the resulting enhanced group of color points. Thus, the resulting enhanced group of color points 301 will be generated such that the first segment is not enhanced to (or beyond) the color gamut, but rather is enhanced to the edge of the neighbor group of color points. Thus, the two segments are not enhanced to the same areas in color space but maintain a difference in the occupied areas of color space. Therefore, color variations will be maintained between the segments and an improved image is achieved. In particular, a perceived loss of detail may effectively be mitigated.

The described approach thus provides an enhancement (saturation increase) in chromaticity while ensuring that no input pixels are mapped to the same color output pixel.

The approach may further be illustrated by FIG. 4 which illustrates color space regions (in the specific example chromaticity areas) for different segments in the input image. The color enhancement is in the specific example a saturation enhancement which maps the input image chromaticity regions to saturation increased regions as illustrated by the mappings F1..F4. These mappings may simply be coordinated as a multiplicative saturation factor, or general non-linear mapping, but in the example F3 and F4 are close in the input image and are therefore also controlled to be close yet not to have any overlap in the output image.

In the example of FIG. 1, the controller comprises an analyzer 113 which is arranged to determine neighboring segments. In the specific example, the analyzer 113 is arranged to consider a first segment which is to be enhanced and to determine one or more neighbor segments for the first segment in accordance with a neighborhood criterion. It will be appreciated that whereas the description focuses on the enhancement of one segment based on considerations of one neighbor segment, the approach may be applied to a plurality and possible all segments of the image and that the enhancement for each segment may consider a plurality of neighbor segments sequentially and/or in parallel. For example, each segment may be enhanced in turn and for each segment to be enhanced, the enhancement may be adopted so that there are no color space overlaps between the color points of the enhanced segment and the color points of any previously enhanced neighbor segment.

It will be appreciated that any suitable neighborhood criterion may be used without detracting from the invention.

In some embodiments, the neighborhood criterion may comprise a requirement that the first segment and the neighbor segment belong to a same image object. In many embodiments, an over segmentation may be used such that relatively small segments are generated. These may then be combined into larger structures that correspond to different image objects. Thus, the image may comprise a number of image objects each of which comprise one or more segments. Indeed, in some cases the image object may comprise a large number of smaller image segments. Each image object may be intended to correspond to a physical object in the image. For example, a house may correspond to one image object, the lawn in front of the house to another image object etc. In such cases, the described approach may be used to ensure that all segments within each image object are enhanced such that they maintain desired inter-segment relationships in color space. Typically, color relationships are important within an image object but less important between image objects and this approach may accordingly provide an improved perceived quality of the color enhancement without unnecessarily restricting this.

In some embodiments, the neighborhood criterion may comprise a requirement that the distance between a point of the first segment in the image and a point of the second segment in the image is less than a threshold. The image points may e.g. be the centers of the segment which may for example be determined as the average spatial position for the pixels of the segment. It will be appreciated that many different approaches for selecting the two appropriate image points may be used. For example, the requirement may correspond to a closest distance (i.e. the distance between the closest points of the two segments) being less than a threshold. As another example, the requirement may correspond to a maximum distance (i.e. the distance between the furthest points of the two segments) being less than a threshold.

The threshold may e.g. be a fixed distance corresponding to an absolute number of pixels or may e.g. be given as a relative distance relative to the image size. In some embodiments, the threshold may be dependent on characteristics of the segments. For example, the distance may be relative to a size of the segments.

As a specific example, it may be determined that two segments are neighbor segments if the minimum distance between them is less than, say, 1/N^th of the image diagonal where N may be e.g. 5, 10, 20 or 40.

In many scenarios a suitable distance may be determined to reflect that segments are neighbor segments if the colors of the segment from a nominal viewing distance are relatable to a human viewer via retinal or brain processing. Indeed, many color models exist that reflect the psycho-visual fact that colors are not perceived individually and in isolation but rather are dependent on neighboring colors and indeed may even depend on surrounding colors e.g. being further away but possible covering larger areas. The distance used to determine whether segments are considered neighbor segments may take into account such psycho-visual effects and may specifically consider segments to be neighbor segments when the color of one segment will have a sufficiently high psycho-visual effect on the perception of the other segment.

Such an approach may allow a practical and low complexity determination of neighbor segments. Furthermore, the determination may provide a good reflection of how likely the two segments are to be perceived relative to each other by a viewer, and thus how important it is to adjust the enhancement based on relative color space characteristics between the two segments.

In some embodiments, the analyzer 113 may be arranged to determine an outlinefor each of the image segments. This outline is determined in the image and may in some embodiments be an inherent characteristic of the segmentation. For example, if the segmenter 103 applies a segmentation algorithm that determines segments as contiguous image areas, the outlines will be provided automatically by the segmenter 103. However, in embodiments where the segmentation may result in non-contiguous areas, the analyzer 113 may proceed to determine an outline using a suitable algorithm. For example, the outline may be determined as an area that meets a given geographical restriction (for example the smallest possible circumference or a having a specific shape, such as a rectangle, circle or ellipse) and which includes a certain percentage of the image points of the segment. As another example, the outline of a segment may be determined as the smallest linear geometric shape that includes all image points.

The criterion for segments being neighbor segments may then comprise a requirement that the first segment and the second segment have less than a threshold of intervening outlines. Specifically, it may be required that there are no other segment outlines between the two segments than those of the segments themselves. This may determine segments to be neighbors if they are adjacent. In some embodiments, it may be allow for e.g. one or two (or possibly more) other outlines to be between the two segments.

As an example, a center point may be determined for each segment and a line drawn from the center point for the segment being considered to each of the other center points. Any segment that does not have any other intervening outlines along the line connecting the segments centers will be considered to be neighbor segments.

The approach may allow an efficient and low complexity determination of suitable neighbor segments that may advantageously be considered together when performing color enhancement. It will be appreciated that the different requirements may be combined.

In some embodiments, the control of the color enhancement may be based on a determination of outlines (or envelopes) for the segments in color space. For example, the color enhancement apparatus of FIG. 1 may comprise an outline processor 115 which is arranged to determine outlines for groups of color points in the color plane. Thus, the outline processor 115 may e.g. be arranged to determine an outline for a group of color points corresponding to the first and neighbor segments following an enhancement. It will be appreciated that the principles described with reference to determination of outlines in the image may also be applied for determination of outlines in color space, and that the principles may easily be extended to determination of three dimensional outlines.

An outline may be an N-dimensional (where N is the number of color space dimensions considered) border of an N dimensional color space region. The outline may define the border of the region considered to be taken up by the group. In many,

embodiments, the outline may e.g. correspond to the peak envelope, i.e. may be the region made by a suitable N-dimensional polygon that contains all color points of the group.

However, in many embodiments, a more complex criterion for defining the outline may be used in order to provide a simpler outline and/or to allow some color points to fall outside the outline.

The outline processor 115 is coupled to the adjuster 107 and is arranged to receive information on color enhanced color point groups and in return to provide the outlines of the color point groups.

In this example, the adjuster 107 may then be arranged to adjust the color enhancement of the first segment based on the outlines of the first segment and the neighbor segment(s). Specifically, the adjuster may be arranged to prevent any overlap between the outlines. This may ensure that the different segments do not cover the same color space regions after color enhancement and may accordingly ensure that the color differentiation between the segments is maintained after the color enhancement.

In some embodiments, the color enhancement apparatus may apply a recursive approach to the adjustment of the color enhancement for the first segment. In such an embodiment, the color enhancement algorithm may first be applied to the first segment using a first value of the characteristic/ parameter to be adjusted. This first value may e.g. be a default value, such as a default color saturation increment, or may e.g. be the value that was used for the previous segment. The adjuster 107 may then proceed to determine the resulting group of color points that originate from the enhanced first segment. The adjuster 107 may then proceed to check whether this resulting group is in conflict with the neighbor group of color points that may specifically correspond to the color points that originate from a color enhancement of the neighbor segment, i.e. the neighbor segment may already have been enhanced. As a specific example, the adjuster 107 may detect whether any unacceptable color space overlap occurs between the two groups.

If no conflict is detected, the already applied color enhancement of the first segment is maintained. However, if a conflict is detected, the adjuster 107 determines a new adjusted characteristic/parameter for the color enhancement and proceeds to perform a new color enhancement of the first segment using the adjusted characteristic. Specifically, a reduced degree of saturation increase may be applied. The process may then be repeated until no conflict is detected.

As mentioned previously the color enhancement apparatus may be arranged to process a plurality and specifically all of the segments of the image. In accordance with some embodiments, the color enhancement apparatus may first determine the image segments and may then order these in order of saturation.

Specifically, the segments are in this example sorted according to their initial average chromaticity and they are then processed in descending order in terms of e.g. their distance to the gamut limit.

First the most saturated segment is processed resulting in an enhanced segment having a corresponding group of color points. As this is the first segment, this enhancement may be performed without consideration of other segments. However, in the case of saturation increase, the enhancement may take into account the available gamut and may in particular perform an enhancement that places the corresponding group of enhanced color points further towards the gamut limit/ edge. Specifically, it may perform the color enhancement such that this group is moved to be proximal to the gamut edge. Thus, the color enhancement apparatus may apply a color enhancement to the segment such that the corresponding group is closer and possibly proximal to the gamut limit. An example is shown in FIG. 2 and 3 where segment 203 is the most saturated and this is shifted to a position 301 in color space which is next to the gamut limit 205.

In the example, the color enhancement apparatus determines an outline for each color enhanced region. Specifically, the space occupied by the group in color space is estimated by determining the minimum (e.g. 15^th percentile) and maximum (e.g. 85^th percentile) values on each axis. This results in each group of color points being considered to correspond to a three dimensional box in a three dimensional color space, and to a two dimensional rectangle in a two dimensional chromaticity plane color space.

The color enhancement apparatus then proceeds to the next most saturated segment. It then evaluates neighbors for this segment and if none are found, it proceeds to perform a default enhancement for the initial segment, i.e. the enhancement is performed without considering any previously enhanced segments. This is repeated until a segment having an already processed neighbor segment is found. In the example, this occurs for the image segment which is a neighbor to the first enhanced segment, and which has the color space group 201 in FIG. 2. Thus, the first enhanced segment becomes a neighbor segment for the current segment (which thus corresponds to the first segment in accordance with the previously used terminology).

The color enhancement apparatus then proceeds to apply a default saturation increase to the first segment and to determine the outline box of the resulting color points. It then checks whether this box overlaps the box of the neighbor segment. If not, the color enhancement apparatus moves on to the next segment but if an overlap is detected, the adjuster 107 proceeds to change the characteristic, and specifically to reduce the saturation increase until no overlap occurs.

The approach is then repeated for the remaining segments. It will be appreciated that for segments that have a plurality of previously processed neighbors, the saturation increase may reduced until the resulting outline box does not overlap the outline box of any of the previously processed neighbors.

Thus, whereas the initial segment can be freely enhanced without considering any other segments, the subsequent segments are first processed to identify any neighbors. If any neighbors are found the enhancement is checked to avoid overlaps in lightness and hue angle against already enhanced segments. In case of overlap, the segment chromaticity shift is reduced until the enhancement does not result in an overlap.

Thus, in such embodiments, the color enhancement apparatus is arranged to sequentially apply the color enhancement algorithm to the image segments in order of decreasing saturation. Furthermore, the adjuster 107 is arranged to adjust the characteristic in response to the outline of the first segment and the outline of the neighbor segment where the neighbor segment is a previously processed image segment.

In the example, the neighbor segment is thus closer to a gamut limit than the first segment and indeed the color enhancement apparatus may apply the color enhancement to the neighbor segment such that this is close to the gamut limit (in accordance with any suitable criterion for closeness and without at this stage considering the neighbor segment as a neighbor segment). The first segment may then be enhanced such that it is in color space moved towards the gamut limit while still ensuring that the overlap is less than a threshold, and specifically in many embodiments that no overlap occurs. It will be appreciated that this approach may be applied in many other examples and is not restricted to the described sequential evaluation in order of decreasing saturation. An example for two segments is illustrated in FIG. 2 and 3.

It will be appreciated that many other adjustments or considerations are possible. For example, the adaptation of the enhancement process may be dependent on other segments which have fairly different colors. Indeed, the human color perception is known to be a complex system that includes a significant amount of relative perception, i.e. the perception of a color of an object is not only based on the actual color of the object but also on colors of other objects. Indeed, color models tend to consider both the object and the proximal areas in which the object is located (referred to as the background). Indeed, the color perception may even be affected by further remote areas (referred to as the surround). Therefore, the current approach may modify the enhancement of a segment based on segments that belong to the background or surround for the image object that the segment belongs to.

For example, if a tomato is placed on lettuce, the green background will result in the tomato appearing redder than if it was placed on e.g. a gray background. The enhancement of the lettuce may result in the tomato being perceived to be redder than prior to the enhancement. This may be compensated by adapting the enhancement of the segments forming the tomato by taking into account the characteristics of the segments forming the lettuce.

In some embodiments, the adjuster 107 may further be arranged to adjust the color enhancement in response to a relative property of the color points for the first segment itself. Thus, the color enhancement may not only be based on the properties of color points for other segments but may also be customized for the specific internal color space characteristics of the segment itself. For example, if the original segment has a gradual color variation (e.g. corresponding to a color gradient), the color enhancement may be adapted to ensure that the color gradient is substantially maintained in the enhanced segment. For example, the color enhancement may allow the gradient to be compressed or expanded and/or moved to different color space areas (specifically increased saturation). However, the gradient shape is maintained to ensure a similar visual appearance. In particular, the color enhancement may be adapted to ensure that the gradient is not distorted by a clipping of the color points within the segment (e.g. even if the available color space region is reduced to avoid overlap with neighbor segments). However, if the segment corresponds to a more random pattern (e.g. a relatively random texture), increased clipping may be allowed as this becomes less perceptible and may allow a larger color variation within the enhanced segment.

The adjuster 107 may thus be arranged to not only consider inter-segment color space characteristics but to also consider intra-segment color space characteristics. For example, it is desirable when transforming (by the color enhancement) a color saturated profile in an image, not just to take into account the pixel values themselves but also their relation to other pixels in the segment, i.e. to consider other spatio-colorimetric properties. E.g. objects may be saturation increased which may improve the perceived image quality. However, the human visual system will evaluate the segments/ pixels in their surroundings and thus relative to other segments and pixels. Such considerations may be particularly important within object profiles (such as e.g. a gradient). In the color enhancement apparatus of FIG. 1, the color enhancement may particularly be controlled to apply a transformation that keeps this intra-segment spatial-colorimetric structure similar (e.g. if the segments has a substantially sigmoid profile, the color enhancement is controlled such that the saturation increased segment also has a sigmoid profile). Indeed, in some embodiments, the color enhancement may be arranged to map each part of the curves to a same psycho-visual appearance. Indeed, the adjuster may be arranged to prevent or reduce color profile changing transformations that significantly change the spatio-colorimetric shape. This may in particular occur for a constant level clipping.

The color enhancement apparatus may in particular achieve this by considering the internal spatio-colorimetric profile of the segment. Specifically, the adjuster may determine a maximum saturation increase that will still allow the profile shape to be maintained. This maximum saturation will thus avoid clipping but may allow the profile to be e.g. compressed. For example, a color gradient may be maintained but made sharper.

In some embodiments, the color enhancement for one segment may be dependent on the spatio-colorimetric profile of a neigbor segment. Thus, the enhancement of a segment may take into account the internal characteristics of another segment. For example, a larger spatial gradient in one segment may be taken into account when determining the gradient in another segment, if these segments are likely to be related in the perception of the viewer.

This may effectively be used to perform a correlation between segments that may provide an improved image. For example, graphics/text may be synchronized to some objects or different objects that are perceived together by a human may maintain correlated characteristics. For example, the enhancement of various segments corresponding to human skin may be correlated even if these are not adjacent. E.g. a segment of a persons shoulder may be enhanced consistently with e.g. the enhancement of a segment of a persons face or arms etc. Such correlated enhancement will allow the perception of the human's skin and whole person to be more natural. The approach may be particularly suitable for segments that are considered neighbors by virtue of belonging to the same object. This skin-colored region may be processed as a whole in several ways, e.g. specular reflections on it may be saturation-increased differently (more excessively) to mitigate this artifact (which is usually excessive for cameras with strong gamma behavior). Furthermore, the perception of color depends not only on the color of the segment itself but also of close/adjacent segments (typically referred to as background in color models) as well as often larger and further away areas (typically referred to as surround). Thus, by taking such features into account when adapting the enhancement of a segment, improved perceived quality can be obtained.

In many embodiments, the enhancement may be depend on both the spatio- colorimetric profile of the segment itself as well as on the spatio-colorimetric profile of another segment. The segments may specifically be segments of the same image object.

An example will be described with reference to FIG. 5 which shows an example of a chromaticity value (xy) as a function of a spatial position (x). For simplicity, both the spatial position and chromaticity values are illustrated as one dimensional parameters so that the example can be described with reference to a 2 dimensional coordinate system.

The example illustrates an initial chromaticity value variation 501 as a function of the spatial position for a first segment prior to enhancement. The example further illustrates an initial chromaticity value variation 503 as a function of the spatial position for a second segment prior to enhancement. The two segments belong to the same image object and are considered to be neighbor segments. However, the two segments are not adjacent but are divided by a third segment 505 which belongs to another image object.

As a specific example, the first and second segments may belong to a red apple whereas the dividing segment may be a darker candle in front of the apple. In this case, it is desirable that the enhancement of the two segments belonging to the apple is correlated such that the close relationship before the enhancement is maintained after.

FIG. 5 illustrates a resulting enhanced chromaticity value variation 507, 509 for the first segment and for the second segment. The enhanced values are generated based on the initial chromaticity value variations 501, 503 for both segments. This is specifically used to ensure that the gradients of the two segments not only reflect the gradients of the non- enhanced segments but also that the enhanced segments are correlated such that the gradients are perceived to corresponds to sections of the same gradient, i.e. that they are correlated to provide a smooth and consistent transition despite the intervening segment of another object.

In practice, this will result in the apple looking more natural and that the two segments of the apple do not exhibit an unnatural step in chromaticity.

Thus, the colors of an object may e.g. span an ellipse in color space between the values (huel,satl) and (hue2, sat2). However, how fast, or erratic up-and-down this function varies may also have an impact on how the neighboring regions are processed. E.g. a red apple illuminated with slightly greenish light and covered in the middle by a pinkish candle (which vertically stretches all along a small strip of the apple, effectively covering it) may be subject to a tripartite coordination. For example, the left side may be dark resulting in the saturated red colors of the apple coming through. However, the right side,may be specularly reflecting, and the desaturating colors of the greenish (anti-red ) illumination may be mixed (e.g. in a Lambertian or Phong, profile). In the middle where the candle is blocking the apple, it may be desirable to apply different coordinations (e.g. to show more clearly that it is a different object than the apple)

Between the two parts of the apple, the dark profile will determine not only where the candle-blocked continuation continuous as to its expected color properties, but it may also be desirable to adapt the spatial slope of that segment (taking into account as calculation constraints a mixture of how much color space is available for transforming and spatio-visual color impression (function of sharpness etc.)).

However for the candle, which is a different image object, it may be desirable to anticorrelate/separate the object from the apple (this may especially be desirable if the apple has a red surround and the color-space chromaticities become so close that the apple starts to melt into its surroundings).

It will be appreciated that many variations and adaptations taking these considerations into account can be thought of. For example, the correlated adaptation may be used to reflect psycho-visual characteristics and compensate or mitigate situations where e.g. the enhancement of one segment undesirably affects the color perception of another segment. Thus, it may specifically consider the relative color perception of human vision.

In some embodiments, the adjuster may also be arranged to maintain a geometric shape of the color space point groups when performing the enhancement.

However, in other embodiments, it may be more desirable to allow the shape to be modified as this provides additional freedom in optimizing the enhancement.

The described approach focused on a sequential color enhancement adjustment where the color enhancement of the neighbor segment is not dependent on the enhancement of the first segment. However, it will be appreciated that in some embodiments the adjuster 107 may also adjust a characteristic of the color enhancement for the neighbor segment in response to the relative color space property between the two. Indeed, the color enhancement of both segments may be jointly modified to result in the desired properties. For example, if an overlap is detected between two segments, the adjuster may proceed to not only reduce the saturation increase for one of the segments but also to increase it for the other segment. For example, the saturation increase may for each iteration be reduced by a predetermined amount for the first segment and increased by a predetermined amount for the second segment until no conflict or overlap in the outline boxes is detected.

It will be appreciated that in some embodiments the determination of the relative property between the two segments may be based on considerations in three (or multidimensional color space. However, in some embodiments, the operation may be based entirely on considerations in the chromaticity plane as exemplified by FIGs. 2 and 3. This may provide a particularly advantageous implementation in many scenarios as the reduced complexity of considering only two dimensions is particularly suitable for color enhancement algorithms that can be sufficiently accurately evaluated in the chromaticity plane. Indeed, the comments to three dimensional color space characteristics may be reduced to two

dimensional considerations while maintaining the same principles. This may substantially reduce complexity and computational burden in many embodiments.

It will be appreciated that the above description for clarity has described embodiments of the invention with reference to different functional circuits, units and processors. However, it will be apparent that any suitable distribution of functionality between different functional circuits, units or processors may be used without detracting from the invention. For example, functionality illustrated to be performed by separate processors or controllers may be performed by the same processor or controllers. Hence, references to specific functional units or circuits are only to be seen as references to suitable means for providing the described functionality rather than indicative of a strict logical or physical structure or organization.

The invention can be implemented in any suitable form including hardware, software, firmware or any combination of these. The invention may optionally be

implemented at least partly as computer software running on one or more data processors and/or digital signal processors. The elements and components of an embodiment of the invention may be physically, functionally and logically implemented in any suitable way. Indeed the functionality may be implemented in a single unit, in a plurality of units or as part of other functional units. As such, the invention may be implemented in a single unit or may be physically and functionally distributed between different units, circuits and processors.

Although the present invention has been described in connection with some embodiments, it is not intended to be limited to the specific form set forth herein. Rather, the scope of the present invention is limited only by the accompanying claims. Additionally, although a feature may appear to be described in connection with particular embodiments, one skilled in the art would recognize that various features of the described embodiments may be combined in accordance with the invention. In the claims, the term comprising does not exclude the presence of other elements or steps.

Furthermore, although individually listed, a plurality of means, elements, circuits or method steps may be implemented by e.g. a single circuit, unit or processor. Additionally, although individual features may be included in different claims, these may possibly be advantageously combined, and the inclusion in different claims does not imply that a combination of features is not feasible and/or advantageous. Also the inclusion of a feature in one category of claims does not imply a limitation to this category but rather indicates that the feature is equally applicable to other claim categories as appropriate.

Furthermore, the order of features in the claims do not imply any specific order in which the features must be worked and in particular the order of individual steps in a method claim does not imply that the steps must be performed in this order. Rather, the steps may be performed in any suitable order. In addition, singular references do not exclude a plurality. Thus references to "a", "an", "first", "second" etc do not preclude a plurality. Reference signs in the claims are provided merely as a clarifying example shall not be construed as limiting the scope of the claims in any way.

Claims

CLAIMS:

1. An apparatus for performing a color enhancement of an image, the apparatus comprising:

a segmenter (103) for generating image segments for the image; an analyzer (113) for identifying at least one neighbor segment of a first segment in accordance with a neighborhood criterion;

a color enhancer (105) for applying a color enhancement algorithm to the first segment;

a processor for determining a resulting group of color points in a color space comprising color points for at least some color enhanced pixels of the first segment;

a processor for determining a neighbor group of color points comprising color points for at least some pixels of the at least one neighbor segment; and

an adjuster (107) for adjusting a characteristic of the color enhancement algorithm for the first segment in response to a geometric property of a distribution of color points of the resulting group of color points relative to a geometric property of a distribution of color points of the neighbor group of color points.

2. The apparatus of claim 1 wherein the color points of the neighbor group of color points correspond to color enhanced pixels.

3. The apparatus of claim 1 or 2 wherein the adjuster (107) is arranged to control the characteristic to restrict an overlap of color points between the first segment and the neighbor segment to be less than a threshold.

4. The apparatus of claim 1 or 2 or 3 further comprising a processor for determining a first outline for at least the resulting group of color points and a second outline for the neighbor group of color points, and wherein the adjuster (107) is arranged to control the characteristic to avoid an overlap between the first outline and the second outline.

5. The apparatus of claim 1, 2, 3 or 4 wherein the neighbor segment is closer to a gamut limit than the first segment and the color enhancer (105) is arranged to apply a color enhancement to the neighbor segment to move this in the color space towards the gamut limit; and the adjuster (107) is arranged to determine the characteristic such that the first segment is in the color space moved towards the gamut limit while restricting an overlap between the resulting group of color points and the neighbor group of color points below a threshold.

6. The apparatus of claim 1 arranged to apply the color enhancement algorithm to the first segment with the characteristic having a first chromaticity modification value, and wherein the adjuster (107) is arranged to determine a modified chromaticity modification value for the characteristic in response to a detection of a conflict between the neighbor group and the resulting group for the first chromaticity modification value, and wherein the color enhancer (105) is arranged to redo the color enhancement by applying the color enhancement algorithm to the first segment using the modified chromaticity modification value.

7. The apparatus of claim 1 wherein the color enhancer (105) is further arranged to apply the color enhancement algorithm to the neighbor segment to generate the neighbor group; and

the adjuster (107) is arranged to adjust a further characteristic of the color enhancement algorithm for the neighbor segment in response to the geometric property of the distribution of color points of the neighbor group of color points relative to the geometric property of the distribution of color points of the resulting group of color points.

8. The apparatus of claim 1 further comprising an outline processor for determining an outline for each of the image segments in the color space, an order processor for ordering the image segments in order of saturation, and a controller for sequentially applying the color enhancement algorithm to the image segments in order of decreasing saturation, and wherein the adjuster (107) is arranged to adjust the characteristic in response to the outline of the first segment and the outline of the neighbor segment where the neighbor segment is a previously processed image segment.

9. The apparatus of claim 1, 3 or 4 wherein the neighborhood criterion comprises a requirement that the first segment and the neighbor segment belong to a same image object.

10. The apparatus of claim 1, 3 or 4 further comprising a processor for

determining an outline in the image for each of the image segments, and wherein the neighborhood criterion comprises a requirement that the first segment and the second segment have less than a threshold of intervening outlines.

11. The apparatus of claim 1 , 3 or 4 wherein the neighborhood criterion comprises a requirement that a distance between an image point of the first segment and an image point of the segment is less than a threshold.

12. The apparatus of claim 1 wherein the adjuster (107) is further arranged to adjust the characteristic in response to relative characteristics for the color points of the resulting group.

13. The apparatus of claim 1, 3 or 4 wherein the adjuster (107) is further arranged to adjust the characteristic in response to a spatio-colorimetric profile of the first segment.

14. The apparatus of claim 1, 3, 4 or 13 wherein the adjuster (107) is further arranged to adjust the characteristic in response to a spatio-colorimetric profile of the neighbor segment.

15. The apparatus of claim 1 wherein the color space is a chromaticity plane.

16. The apparatus of claim 1 wherein the color enhancement algorithm is a saturation enhancement algorithm.

17. A method of color enhancement of an image, the method comprising:

generating image segments for the image;

identifying at least one neighbor segment of a first segment in accordance with a neighborhood criterion;

applying a color enhancement algorithm to the first segment;

determining a resulting group of color points in a color space comprising color points for at least some color enhanced pixels of the first segment;

determining a neighbor group of color points comprising color points for at least some pixels of the at least one neighbor segment; and