| With the rapid development of big data on the Internet,people pay more and more attention to the in-depth mining of valuable information in images,but how to quickly and efficiently extract key information in images and solve new problems brought by data redundancy.Salient region detection can quickly extract key information in the image,effectively solve redundant information in the image,reduce the amount of calculation,and is a key step in computer vision tasks.Therefore,the salient region detection algorithm is widely used in the field of computer vision and has achieved significant results.Although many salient region detection algorithms have been proposed at present,the detection results are still unsatisfactory,and the background noise is not well suppressed,and the range of the salient area is not complete,and the detection results are not ideal.These problems have caused great errors in the precision of applied research in the field of computer vision.Existing these problems,two salient region detection algorithms are proposed in this paper.One of them is Multi-scale Saliency Detection Based on Bayesian Framework.Another is Saliency Detection Based on Multi-priorities Fusion.The main research work of this paper includes:(1)A multi-scale Bayesian based saliency detection algorithm is proposed to improve the unsatisfactory accuracy of traditional Bayesian based saliency detection methods.Firstly,multi-scale super-pixels are generated by segmenting the input image with super-pixel segmentation algorithm(SLIC).The background seeds are obtained according to the boundary information of super-pixels,followed by the background prior evaluation with distance computation and multi-scale fusion.Secondly,the Harris operator is used to detect the corner points of the enhanced image to obtain the convex hulls.Multi-scale super-pixels are fused and result in a convex hull prior.Then,the final prior is generated by combining the background prior and convex hull prior.Meanwhile,the observation likelihood probability is computed by using the color histogram.Finally,the saliency map is evaluated with Bayesian model according to the obtained prior probability map and observation likelihood probability.Comparing the accuracy and recall rate with various algorithms on the data sets MSRA1000,ECSSD and DUT-OMRON,the algorithm has better performance.(2)This paper proposes an improved algorithm based on multi-priorities fusion to improve less effective in detecting images with complex background saliency detection algorithm.Firstly,the image is divided into four different super-pixel scales,and the boundary saliency maps of different scales are calculated according to the MR algorithm,and linear fusion is performed to obtain the final boundary prior.Secondly,the background seeds are obtained according to the boundary information,followed by the background prior evaluation with distance and Cellular Automata.Then,consider the visual system to pay more attention to the warm color,while the warm tone has an effect on the image saliency,thus adding color prior.Finally,the saliency map is evaluated with Multi-Cellular Automata model according to the obtained three prior probability map.Experimental comparison with various algorithms on the data sets MSRA1000,ECSSD and DUT-OMRON,the algorithm has better performance.(3)In this paper,the proposed algorithms are analyzed experimentally on standard data sets MSRA1000,ECSSD and DUT-OMRON.Compare the two saliency algorithms proposed in this paper with 12 mainstream saliency detection algorithms.Through experimental analysis and comparison,it is found that the algorithm in this paper is better than the comparison algorithm,the visual effect is better,and the accuracy is significantly improved.At the same time,the saliency detection algorithm in this paper is applied to the actual scene of marine target detection.Experiments show that the algorithm in this paper can effectively improve the detection accuracy of salient region and obtain better detection results. |
PDF Full Text Request