Content Based Image Retrieval Using Local Directional Pattern and Color Histogram

The main focus of content based image retrieval research is to develop feature extraction method in terms of expressing effective texture, color and shape features in a similar way of human visual perception. In this chapter, a new feature extraction method is developed by using Local Directional Pattern (LDP) and color histogram which not only can capture color, texture and shape properties, but also utilize different color spaces effectively. First, RGB image is converted into HSV model, and LDP descriptor is used to describe visual texture and geometrical features using the V (value) image in HSV color space. Then color texture is extracted from color histogram in RGB color space with color quantization. Finally these features are fused in final stage for image retrieval with different distance metrics. The WANG image database is used to validate the proposed method effectively, and the results demonstrate that the proposed approach is more effective for image retrieval and can be used directly on natural images without any segmentation and preprocessing.