Generalized Rotation-gray Element Co-occurrence Matrix Based Optimization for CBIR in Web-based Mini-PACS
Xiaocao Cao, Xubo Yang, Lixu Gu, Ying Yuan, Mingting Yu · 2009
Since the algorithm based on gray level co-occurrence matrix does not have scale invariance and rotation invariance, a new concept called generalized grayscale pixel rotation is proposed in this paper. This concept involves both resizing factor and rotation factor of the image. On that basis, a new algorithm called GRECM (Generalized Rotation-gray Element Co-occurrence matrix)-based algorithm is proposed for optimizing GLCM-based algorithm. By applying this algorithm as one of the CBIR (Content-Based Image Retrieval) algorithms to the mini-PACS (Picture Archiving and Communication System) platform, ideal experimental results have been obtained. Compared with the traditional algorithm based on GLCM (ray Level Co-occurrence Matric) or GGLCM (Generalized Gray Level Co-occurrence Matrix), the new algorithm largely reduces its sensitivity to the scale of the image or the direction of the object on the image. So this algorithm has very good performance in improving the retrieval precision when there are a lot of images resized or rotated by the same image in the image library.