作者: Callins Christiyana Chelladurai , Rajamani Vayanaperumal
DOI: 10.17485/IJST/2015/V8I7/62845
关键词: Feature (computer vision) 、 Content-based image retrieval 、 Pixel 、 Artificial intelligence 、 Computer vision 、 Pattern recognition 、 Image retrieval 、 Computer science 、 Precision and recall 、 Pattern recognition (psychology) 、 Feature extraction 、 Radius
摘要: Objectives: This work proposes a feature extraction procedure named as Global Neighbour Preserving Local Ternary Co-occurrence Pattern (GNPLTCoP) in the Content Based Image Retrieval (CBIR) Task for ultrasound kidney images retrieval. Methods/Analysis: The proposed GNPLTCoP finds local pattern based on co-occurrence of first order derivatives ternary fashion from radius 1 and 2 neighbourhoods center pixel small 3X3 square region. Then is classified into two to integrate global information. classification correlation between region mean intensities. Findings: Performance compared with (LTCoP) it computes derivatives. LTCoP considers 8 pixels neighbourhood candidates neighbourhood. Among 16 neighbourhood, are used information left computation. problem addressed computation GNPLTCoP. different by means considering adding replaced value itself its neighbours preserve neighbour employs image retrieval system consists database images. performance LTP LTCoP. discriminative power substantiated through Precision Recall measures. Conclusion/Application: can be applied other types medical recognition applications.