While image retrieval and image compression have been pursued separately in the past, compressed domain techniques, which allow processing or retrieval of images without prior decompression, are becoming increasingly important. In this chapter we show that such midstream content access is possible and present a compressed domain retrieval method based on a visual pattern based compression algorithm. Experiments conducted on a medium sized image database demonstrate the effectiveness and efficiency of the presented approach.
Colour Visual Pattern Image Coding (Cvpic)
The Colour Visual Pattern Image Coding (CVPIC) image compression algorithm introduced in (Schaefer et al., 1999) is an extension of the work in (Chen & Bovic, 1990). The underlying idea is that within a 4 × 4 image block only one discontinuity is visually perceptible.
CVPIC first performs a conversion to the CIEL*a*b* colour space (CIE, 1986) as a more appropriate image representation. As many other colour spaces, CIEL*a*b* comprises one luminance and two chrominance channels. CIEL*a*b* however, was designed to be a uniform representation, meaning that equal differences in the colour space correspond to equal perceptual differences. A quantitative measurement of these colour differences was defined using the Euclidean distance in the L*a*b* space and is given in ΔE units.
A set of 14 patterns of 4 × 4 pixels has been defined in (Chen & Bovic, 1990). All these patterns contain one edge at various orientations (vertical, horizontal, plus and minus 45°) as can be seen in Figure 1 where + and − represent different intensities. In addition a uniform pattern where all intensities are equal is being used.
The 14 edge patterns used in CVPIC
The image is divided into 4 × 4 pixel blocks. Determining which visual pattern represents each block most accurately then follows. For each of the visual patterns the average L*a*b* values for the regions marked by + and − respectively (i.e., the mean values for the regions on each side of the pattern) are calculated.
Key Terms in this Chapter
Content-Based Image Retrieval (CBIR): Retrieval of images based not on keywords or annotations but based on features extracted directly from the image data.
Colour Histogram: A feature often used for CBIR where colour space is quantised and the numbers of pixels that fall within each quantisation bin are stored.
Compressed-Domain Image Retrieval: CBIR performed directly in the compressed domain of images (i.e., without a need to uncompress the images first).
Image Similarity Metric: Quantitative measure whereby the features of two images are compared in order to provide a judgement related to the visual similarity of the images.
Query-by-Example Retrieval: Retrieval paradigm in which a query is provided by the user and the system retrieves instances similar to the query.
Image compression: The process of storing images in a more compact form by removing redundant data and/or discarding visually less important information.