Combined global-local specialized feature descriptor for content based image retrieval under noisy query

The lot of research on content based image retrieval (CBIR) considering the different image / visual features like color, shape, texture and semantic methods has been done earlier. In the real world environment the noises that may embed into an image document will affect the CBIR environments algorithms. Tough different filtering algorithms are available for noise reduction, applying a filtering algorithm that is sensitive to one type of noise to an image which has been degraded by another type of noise lead to unfavorable results. This condition stresses the importance of designing a efficient CBIR algorithm that retains precision rates even under noisy conditions. In this work, numerous experiments have been conducted to analyze the robustness of our proposed Combined Global - Local Specialized Features Descriptor (CGLSFD). This proposed methods include two stages. First apply wavelet to decompose the query image to extracted the energy, standard deviation and mean values in all bands. Second apply micro structure descriptor (MSD) to extract image edge orientation features with color, texture and shape and color layout information. This proposed method extensively tested on Corel data tests. This proposed CGLSFD algorithm has high indexing and low dimensionality also along with other existing conventional algorithms under different types of noises such as Gaussian noise, salt and pepper noise and quantization noises. This proposed CGLSFD algorithms results compare with other existing conventional algorithms than Gabor features and MSD in image retrieval.