Extraction of perceptually important colors and similarity measurement for image matching, retrieval and analysis

Abstract

Color descriptors are among the most important features used in image analysis and retrieval. Due to its compact representation and low complexity, direct histogram comparison is a commonly used technique for measuring the color similarity. However, it has many serious drawbacks, including a high degree of dependency on color codebook design, sensitivity to quantization boundaries, and inefficiency in representing images with few dominant colors. In this paper, we present a new algorithm for color matching that models behavior of the human visual system in capturing color appearance of an image. We first develop a new method for color codebook design in the Lab space. The method is well suited for creating small fixed color codebooks; for image analysis, matching, and retrieval. Then we introduce a statistical technique to extract perceptually relevant colors. We also propose a new color distance measure that is based on the optimal mapping between two sets of color components representing two images. Experiments comparing the new algorithm to some existing techniques show that these novel elements lead to better match to human perception in judging image similarity in terms of color composition.

Compact Signature-Based Compressed Video Matching Using Dominant Color Profiles (DCP)

“…The dataset were tested for video retrieval using sum of squared pixel difference (SSD) measure [19] between videos' keyframes, leaving any temporal information without utilization and relying on annotations which is neither accurate (due to dependency on human element) nor always available. Color proved to be a powerful feature regarding image retrieval [22,23] and video retrieval [24,25], as it's strongly related to semantic similarity [26,27]. Color by nature is invariant to partial occlusion, cropping, translation or affine transformations such as scaling, rotation, shear or reflection [24].…”

“…Color by nature is invariant to partial occlusion, cropping, translation or affine transformations such as scaling, rotation, shear or reflection [24]. Color feature is very powerful and could act as a building block for efficient video matching techniques, especially in absence of any semantic cues [26]. Furthermore, humans tend to see scenes as a set of dominant colors [26,28] as it was found that a small number of colors are sufficient to describe any region instead of a full color space [29].…”

“…Color feature is very powerful and could act as a building block for efficient video matching techniques, especially in absence of any semantic cues [26]. Furthermore, humans tend to see scenes as a set of dominant colors [26,28] as it was found that a small number of colors are sufficient to describe any region instead of a full color space [29]. Those important remarks were taken by researchers to utilize and develop more efficient video retrieval techniques starting by using the popular color histogram extracted from DC-images to act as signature for video retrieval [30,31], or even the more sophisticated histograms such as Dominant Color Histograms (DCH) [32].…”

“…Recently it was used for video summarization [35]. DCH is still a global feature that doesn't encode neither spatial nor temporal information [26]; also it is a color space dependent which makes it not effective for robust video retrieval. Those problems will be covered along with the DCP in next section.…”

Perceptually-tuned multiscale color-texture segmentation

“…. Finally, a perceptual metric (OCCD) [12] is used to determine the similarity of two color feature vectors.…”