Saliency Target Detection Model Based On Frequency Domain Analysis

With the development of science and technology,mobile phones,computers and other electronic devices are closely related to people's lives.In daily life,the information received is exponential growth,and the visual information is the main part.In the face of complex and changeable information,how to deal with the information of human visual system has become a hot topic.In addition,the concept of computer vision is put forward,which is one of the most important research directions.As we all know,feature extraction is one of the most important steps in image processing.Therefore,in this paper,several salient features of the image are analyzed,and some methods of feature extraction are introduced.In addition,different color space will have an important impact on the final detection results.In this paper,different color space models are studied.In this paper,the first algorithm is based on the wavelet Pyramid salient target detection model.This model is inspired by Itti's visual attention model.In order to make use of the ability of local information representation and multi-scale analysis,we use the wavelet Pyramid model to replace the Gauss Pyramid model.The wavelet transform can preserve the better edge information of the saliency image.In addition,we integrate the improved prior-center knowledge into the wavelet Pyramid model,and further enhance the detection results of the target,which is beneficial to the saliency target segmentation.The second model established is based on local analysis and global analysis of images in this paper.Because of its excellent performance,the wavelet transform of the image can be used to analyze the extracted feature images and obtain the local feature.However,because of its ability to express local details,it is easy to detect the incomplete of the target and the background information of complex texture.Therefore,the article integrates the spectral residual algorithm to make a global analysis of the image and get the saliency image of the global analysis.The two salient features are fused to get the final image.The experimental results show that this model is better than other algorithms.The last model of this paper is based on the analysis of image frequency domain.In fact,whether it is based on the image wavelet transform or spectral residual algorithm of the salient target detection methods,they are based on the image frequency domain analysis of a target detection model.In this paper,a novel saliency detection model based on the frequency domain analysis is proposed,which is based on the frequency domain tuning algorithm and the salient target detection model based on local and global analysis.This model is tested on the MSRA 10K and ECS SD databases.The results show that the proposed model can adapt to different types of images,and the detection results are better than other algorithms in the face of multiple targets,large targets and complex background.