Indexing and retrieval of compound color objects using co-occurrence histograms of color and wavelet features

In this paper, we present a system for the retrieval and indexing of images of compound color objects. Compound color objects are objects that consist of a specific set of colors that are spatially arranged in a unique way. Examples of compound color objects include flags, trademarks, logos, and cartoons. In this paper, we apply our proposed technique to logos and trademarks. We introduces a 5-dimensional cooccurrence histogram that captures color and texture information simultaneously. We use the multi-resolution analysis feature with the Coiflet wavelet to capture the texture information in the image. We call this 5-dimensional histogram the color wavelet co-occurrence histogram (CWCH). We show that the CWCH performs better than the edge gradient histogram (EGH) and the color edge co-occurrence histogram (CECH) and the MPEG-7 edge histogram descriptor (EHD).