An Enhanced Approach of CBIR using Gabor Wavelet and Edge Histogram Descriptor
Diksha Kurchaniya, Punit Kumar Johari · International Journal of Signal Processing Image Processing and Pattern Recognition · 2017
In attention to the last decades with the rapid growth of digital images on the internet and digital image repository, retrieval of these images requires some efficient and effective image retrieval techniques accordance with the user perception.Many researchers have been proposed different image retrieval techniques.Content-Based Image Retrieval (CBIR) provides an effective approach to search and retrieve images from large databases, but to develop a CBIR system with an appropriate combination of low-level features is a big issue.To overcome these issues we used a combination of Gabor Wavelet and Edge Histogram Descriptor.The proposed approach is based on Gabor wavelet filter that have the capacity to collapse the filter reactions as per the scale and orientation of the textures.In Gabor wavelet, wavelets capture energy at a particular frequency and particular direction, we applied Gabor Wavelet filter in four different directions of the image such as-Horizontal, Vertical, Diagonal and Anti-Diagonal.Edges give the object illustration of an image and used as a feature descriptor for image retrieval.Spatial information about five different types of edges of the image is provided by EHD.EHD also gives geometrical information of image if same image has different colors.For calculating distance, Euclidean distance is used and average precision and recall are used for performance evaluation of proposed system.Experiment result shows that proposed method provides the improved precision over the existing method.