A New Laws Filtered Local Binary Pattern Texture Descriptor for Ultrasound Kidney Images Retrieval
Content Based Image Retrieval (CBIR) is an inevitable technique in medical applications. One of the important tasks in CBIR is the feature extraction process. A new feature extraction procedure called Laws Filtered Local Binary Pattern (LFLBP) for extracting texture features from ultrasound kidney images is proposed in this manuscript. This new texture feature combines the gain of Laws Masks and Local Binary Pattern (LBP). The Laws Masks enhance the discrimination power of LBP by capturing high energy texture points in an image and efficiently characterize the textures. The new descriptor is intended to utilize the local information in an effective manner neither the increase of encoding levels nor the usage of adjacent neighbourhood information. The performance of this new descriptor is compared with the LBP and the Local Ternary Pattern (LTP). The experimental results show that the ultrasound kidney images retrieval system with this new descriptor has good average precision value (77%) as compared to LBP (74%) and LTP (74.3%).
Content Based Image Retrieval (CBIR), Texture, Local Patterns, Local Binary Pattern (LBP), Local Ternary Pattern (LTP), Laws Masks, Retrieval efficiency