{"title":"一种改进的编码孔径深度判断方法","authors":"Zhang Yu-peng, Wang Yong-tian, Weng Dong-dong","doi":"10.1109/YCICT.2010.5713121","DOIUrl":null,"url":null,"abstract":"Depth estimation is an important part in computer vision, which can be used in augmented reality and pattern recognition etc. At present, there are many methods of depth extraction, such as depth from defocus (DFD), depth from focus (DFF), stereo vision etc. Usually, depth information is very hard to be extracted from only one single shot. In that case, the calculation is lack of constraint. Coded aperture is a method of computational photography which just modifies the lens to plus some priors to realize the depth computation. We did some research on it, but found some problems in the process of layer judgment in depth extraction with coded aperture from one shot. The depth layer judgment in the existing method calculates the sum of convolution errors and derivatives priors, and determines the depth layer in which the minimum value is. However, this calculation is not very reliable and needs manually adjusting different parameters for different original images. In order to overcome the abovementioned problem, a novel depth judgment method is proposed. We note that canny operator is very sensitive to ringing effect and good to blurred image edge as well. According to the restored image sequence of the coded aperture, we use canny edge detection and morphological algorithm to score for every image, then we can judge which the proper one is. Experiments have proved that the proposed method is simple and effective.","PeriodicalId":179847,"journal":{"name":"2010 IEEE Youth Conference on Information, Computing and Telecommunications","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2010-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"An improved method of depth judgment in coded aperture\",\"authors\":\"Zhang Yu-peng, Wang Yong-tian, Weng Dong-dong\",\"doi\":\"10.1109/YCICT.2010.5713121\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Depth estimation is an important part in computer vision, which can be used in augmented reality and pattern recognition etc. At present, there are many methods of depth extraction, such as depth from defocus (DFD), depth from focus (DFF), stereo vision etc. Usually, depth information is very hard to be extracted from only one single shot. In that case, the calculation is lack of constraint. Coded aperture is a method of computational photography which just modifies the lens to plus some priors to realize the depth computation. We did some research on it, but found some problems in the process of layer judgment in depth extraction with coded aperture from one shot. The depth layer judgment in the existing method calculates the sum of convolution errors and derivatives priors, and determines the depth layer in which the minimum value is. However, this calculation is not very reliable and needs manually adjusting different parameters for different original images. In order to overcome the abovementioned problem, a novel depth judgment method is proposed. We note that canny operator is very sensitive to ringing effect and good to blurred image edge as well. According to the restored image sequence of the coded aperture, we use canny edge detection and morphological algorithm to score for every image, then we can judge which the proper one is. Experiments have proved that the proposed method is simple and effective.\",\"PeriodicalId\":179847,\"journal\":{\"name\":\"2010 IEEE Youth Conference on Information, Computing and Telecommunications\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2010-11-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2010 IEEE Youth Conference on Information, Computing and Telecommunications\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/YCICT.2010.5713121\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2010 IEEE Youth Conference on Information, Computing and Telecommunications","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/YCICT.2010.5713121","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
An improved method of depth judgment in coded aperture
Depth estimation is an important part in computer vision, which can be used in augmented reality and pattern recognition etc. At present, there are many methods of depth extraction, such as depth from defocus (DFD), depth from focus (DFF), stereo vision etc. Usually, depth information is very hard to be extracted from only one single shot. In that case, the calculation is lack of constraint. Coded aperture is a method of computational photography which just modifies the lens to plus some priors to realize the depth computation. We did some research on it, but found some problems in the process of layer judgment in depth extraction with coded aperture from one shot. The depth layer judgment in the existing method calculates the sum of convolution errors and derivatives priors, and determines the depth layer in which the minimum value is. However, this calculation is not very reliable and needs manually adjusting different parameters for different original images. In order to overcome the abovementioned problem, a novel depth judgment method is proposed. We note that canny operator is very sensitive to ringing effect and good to blurred image edge as well. According to the restored image sequence of the coded aperture, we use canny edge detection and morphological algorithm to score for every image, then we can judge which the proper one is. Experiments have proved that the proposed method is simple and effective.