Research On Region Of Interest Extraction

With the rapid development of mobile internet technology,a great deal of digital image information is produced which brings forward higher request to image retrieval and storage.In order to reduce the amount of data processed by the digital image technology and improve operation efficiency,region of interest(ROI)is extracted to replace original image,which contains the main information of image.This paper makes following research on ROI extraction and analysis.Related research work is as follows:Firstly,an image segmentation method for saliency maps is proposed in this paper.The saliency maps are obtained by an improved Stentiford method,which improves a lot in image processing efficiency and effectiveness.The image segmentation is mainly divided to two steps:binary segmentation and skeleton extraction.Experiments show that the method can avoid noise interference and have a desired segmentation effect.Secondly,a large number of image features are extracted from input images,saliency maps and skeleton images,including the saliency entropy,saliency entropy,saliency entropy density and etc.Finally,the association rule algorithm is applied to analyze the image features.In this paper,after the characteristics of color images,saliency maps and skeleton images are extracted,FP-growth association rule algorithm is applied to extract the frequent patterns of features and obtain the relationship between image features and image regions.Experiments show that human eye vision system can be well simulated.The single visual model can't greatly satisfy the visual characteristics in the extraction of image region of interest.Based on the improved visual model,the image features are extracted in this paper and the association rule algorithm is applied to analyze the relationship between the features and regions of interest.Experiments show that the method has a better visual effect.