Kernel regression has been previously proposed as a robust estimator for a wide range of image processing tasks, including image denoising, interpolation and super resolution. In this article we propose a kernel formulation that relaxes the usual symmetric and unimodal properties to effectively exploit the smoothness characteristics of natural images. The proposed method extends the kernel support along similar image characteristics to further increase the robustness of the estimates. Application of the proposed method to image denoising yields significant improvement over the previously reported regression methods and produces results comparable to the state-of the-art denoising techniques.
{"title":"Asymmetric, Non-unimodal Kernel Regression for Image Processing","authors":"Damith J. Mudugamuwa, W. Jia, Xiangjian He","doi":"10.1109/DICTA.2010.34","DOIUrl":"https://doi.org/10.1109/DICTA.2010.34","url":null,"abstract":"Kernel regression has been previously proposed as a robust estimator for a wide range of image processing tasks, including image denoising, interpolation and super resolution. In this article we propose a kernel formulation that relaxes the usual symmetric and unimodal properties to effectively exploit the smoothness characteristics of natural images. The proposed method extends the kernel support along similar image characteristics to further increase the robustness of the estimates. Application of the proposed method to image denoising yields significant improvement over the previously reported regression methods and produces results comparable to the state-of the-art denoising techniques.","PeriodicalId":246460,"journal":{"name":"2010 International Conference on Digital Image Computing: Techniques and Applications","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2010-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128704761","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Silhouettes are common features used by many applications in computer vision. For many of these algorithms to perform optimally, accurately segmenting the objects of interest from the background to extract the silhouettes is essential. Motion segmentation is a popular technique to segment moving objects from the background, however such algorithms can be prone to poor segmentation, particularly in noisy or low contrast conditions. In this paper, the work of [1] combining motion detection with graph cuts, is extended into two novel implementations that aim to allow greater uncertainty in the output of the motion segmentation, providing a less restricted input to the graph cut algorithm. The proposed algorithms are evaluated on a portion of the ETISEO dataset using hand segmented ground truth data, and an improvement in performance over the motion segmentation alone and the baseline system of [1] is shown.
{"title":"Accurate Silhouettes for Surveillance - Improved Motion Segmentation Using Graph Cuts","authors":"Daniel Chen, S. Denman, C. Fookes, S. Sridharan","doi":"10.1109/DICTA.2010.69","DOIUrl":"https://doi.org/10.1109/DICTA.2010.69","url":null,"abstract":"Silhouettes are common features used by many applications in computer vision. For many of these algorithms to perform optimally, accurately segmenting the objects of interest from the background to extract the silhouettes is essential. Motion segmentation is a popular technique to segment moving objects from the background, however such algorithms can be prone to poor segmentation, particularly in noisy or low contrast conditions. In this paper, the work of [1] combining motion detection with graph cuts, is extended into two novel implementations that aim to allow greater uncertainty in the output of the motion segmentation, providing a less restricted input to the graph cut algorithm. The proposed algorithms are evaluated on a portion of the ETISEO dataset using hand segmented ground truth data, and an improvement in performance over the motion segmentation alone and the baseline system of [1] is shown.","PeriodicalId":246460,"journal":{"name":"2010 International Conference on Digital Image Computing: Techniques and Applications","volume":"362 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2010-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115943557","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper compares several camera calibration methods on the estimation of specific extrinsic and intrinsic parameters. Good estimates of the chosen parameters, rotation and radial lens distortion are essential to increase the accuracy of quantitative measurements and to accurately stitch single field-of-view-based images together. The parameters are obtained using two selected methods on different objective magnifications on a microscope system using a fixed grid calibration pattern. We evaluate two methods and show that the rotation angles from one of the methods is consistent with a simple homography while the other estimates a consistently smaller angle. The radial distortion estimates are both very small and relate to a distortion of less than one pixel.
{"title":"On the Estimation of Extrinsic and Intrinsic Parameters of Optical Microscope Calibration","authors":"Doreen Altinay, A. Bradley, A. Mehnert","doi":"10.1109/DICTA.2010.43","DOIUrl":"https://doi.org/10.1109/DICTA.2010.43","url":null,"abstract":"This paper compares several camera calibration methods on the estimation of specific extrinsic and intrinsic parameters. Good estimates of the chosen parameters, rotation and radial lens distortion are essential to increase the accuracy of quantitative measurements and to accurately stitch single field-of-view-based images together. The parameters are obtained using two selected methods on different objective magnifications on a microscope system using a fixed grid calibration pattern. We evaluate two methods and show that the rotation angles from one of the methods is consistent with a simple homography while the other estimates a consistently smaller angle. The radial distortion estimates are both very small and relate to a distortion of less than one pixel.","PeriodicalId":246460,"journal":{"name":"2010 International Conference on Digital Image Computing: Techniques and Applications","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2010-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117218953","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This work describes the attribute evaluation sections of the ambitious goal of creating a large-scale content-based image retrieval (CBIR) system for solar phenomena in NASA images from the Solar Dynamics Observatory mission. This mission, with its Atmospheric Imaging Assembly (AIA), is generating eight 4096 pixels x 4096 pixels images every 10 seconds, leading to a data transmission rate of approximately 700 Gigabytes per day from only the AIA component (the entire mission is expected to be sending about 1.5 Terabytes of data per day, for a minimum of 5 years). We investigate unsupervised and supervised methods of selecting image parameters and their importance from the perspective of distinguishing between different types of solar phenomena by using correlation analysis, and three supervised attribute evaluation methods. By selecting the most relevant image parameters (out of the twelve tested) we expect to be able to save 540 Megabytes per day of storage costs for each parameter that we remove. In addition, we also applied several image filtering algorithms on these images in order to investigate the enhancement of our classification results. We confirm our experimental results by running multiple classifiers for comparative analysis on the selected image parameters and filters.
{"title":"Selection of Image Parameters as the First Step towards Creating a CBIR System for the Solar Dynamics Observatory","authors":"J. Banda, R. Angryk","doi":"10.1109/DICTA.2010.94","DOIUrl":"https://doi.org/10.1109/DICTA.2010.94","url":null,"abstract":"This work describes the attribute evaluation sections of the ambitious goal of creating a large-scale content-based image retrieval (CBIR) system for solar phenomena in NASA images from the Solar Dynamics Observatory mission. This mission, with its Atmospheric Imaging Assembly (AIA), is generating eight 4096 pixels x 4096 pixels images every 10 seconds, leading to a data transmission rate of approximately 700 Gigabytes per day from only the AIA component (the entire mission is expected to be sending about 1.5 Terabytes of data per day, for a minimum of 5 years). We investigate unsupervised and supervised methods of selecting image parameters and their importance from the perspective of distinguishing between different types of solar phenomena by using correlation analysis, and three supervised attribute evaluation methods. By selecting the most relevant image parameters (out of the twelve tested) we expect to be able to save 540 Megabytes per day of storage costs for each parameter that we remove. In addition, we also applied several image filtering algorithms on these images in order to investigate the enhancement of our classification results. We confirm our experimental results by running multiple classifiers for comparative analysis on the selected image parameters and filters.","PeriodicalId":246460,"journal":{"name":"2010 International Conference on Digital Image Computing: Techniques and Applications","volume":"63 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2010-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115415765","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Text data in an image present useful information for annotation, indexing and structuring of images. The gathered information from images can be applied for devices for impaired people, navigation, tourist assistance or georeferencing business. In this paper we propose a novel algorithm for text detection and localization from outdoor/indoor images which is robust against different font size, style, uneven illumination, shadows, highlights, over exposed regions, low contrasted images, specular reflections and many distortions which makes text localization task harder. A binarization algorithm based on difference of gamma correction and morphological reconstruction is realized to extract the connected components of an image. These connected components are classified as text and non test using a Random Forest classifier. After that text regions are localized by a novel merging algorithm for further processing.
{"title":"A Novel Algorithm for Text Detection and Localization in Natural Scene Images","authors":"Sezer Karaoglu, Basura Fernando, A. Trémeau","doi":"10.1109/DICTA.2010.115","DOIUrl":"https://doi.org/10.1109/DICTA.2010.115","url":null,"abstract":"Text data in an image present useful information for annotation, indexing and structuring of images. The gathered information from images can be applied for devices for impaired people, navigation, tourist assistance or georeferencing business. In this paper we propose a novel algorithm for text detection and localization from outdoor/indoor images which is robust against different font size, style, uneven illumination, shadows, highlights, over exposed regions, low contrasted images, specular reflections and many distortions which makes text localization task harder. A binarization algorithm based on difference of gamma correction and morphological reconstruction is realized to extract the connected components of an image. These connected components are classified as text and non test using a Random Forest classifier. After that text regions are localized by a novel merging algorithm for further processing.","PeriodicalId":246460,"journal":{"name":"2010 International Conference on Digital Image Computing: Techniques and Applications","volume":"70 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2010-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114814639","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The computation of stereoscopic depth is an important field of computer vision. Although a large variety of algorithms has been developed, the traditional correlation-based versions of these algorithms are prevalent. This is mainly due to easy implementation and handling but also to the linear computational complexity, as compared to more elaborated algorithms based on diffusion processes, graph-cut or bilateral filtering. In this paper, we introduce a new two-stage matching cost for the traditional approach: the summed normalized cross-correlation (SNCC). This new cost function performs a normalized cross-correlation in the first stage and aggregates the correlation values in a second stage. We show that this new measure can be implemented efficiently and that it leads to a substantial improvement of the performance of the traditional stereo approach because it is less sensitive to high contrast outliers.
{"title":"A Two-Stage Correlation Method for Stereoscopic Depth Estimation","authors":"Nils Einecke, J. Eggert","doi":"10.1109/DICTA.2010.49","DOIUrl":"https://doi.org/10.1109/DICTA.2010.49","url":null,"abstract":"The computation of stereoscopic depth is an important field of computer vision. Although a large variety of algorithms has been developed, the traditional correlation-based versions of these algorithms are prevalent. This is mainly due to easy implementation and handling but also to the linear computational complexity, as compared to more elaborated algorithms based on diffusion processes, graph-cut or bilateral filtering. In this paper, we introduce a new two-stage matching cost for the traditional approach: the summed normalized cross-correlation (SNCC). This new cost function performs a normalized cross-correlation in the first stage and aggregates the correlation values in a second stage. We show that this new measure can be implemented efficiently and that it leads to a substantial improvement of the performance of the traditional stereo approach because it is less sensitive to high contrast outliers.","PeriodicalId":246460,"journal":{"name":"2010 International Conference on Digital Image Computing: Techniques and Applications","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2010-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127245317","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper presents a wavelet-based texture analysis method for classification of melanoma. The method applies tree-structured wavelet transform on different color channels of red, green, blue and luminance of dermoscopy images, and employs various statistical measures and ratios on wavelet coefficients. Feature extraction and a two-stage feature selection method, based on entropy and correlation, were applied to a train set of 103 images. The resultant feature subsets were then fed into four different classifiers: support vector machine, random forest, logistic model tree and hidden naive bayes to classify melanoma in a test set of 102 images, which resulted in an accuracy of 88.24% and ROC area of 0.918. Comparative study carried out in this paper shows that the proposed feature extraction method outperforms three other wavelet-based approaches.
{"title":"Classification of Melanoma Lesions Using Wavelet-Based Texture Analysis","authors":"R. Garnavi, M. Aldeen, J. Bailey","doi":"10.1109/DICTA.2010.22","DOIUrl":"https://doi.org/10.1109/DICTA.2010.22","url":null,"abstract":"This paper presents a wavelet-based texture analysis method for classification of melanoma. The method applies tree-structured wavelet transform on different color channels of red, green, blue and luminance of dermoscopy images, and employs various statistical measures and ratios on wavelet coefficients. Feature extraction and a two-stage feature selection method, based on entropy and correlation, were applied to a train set of 103 images. The resultant feature subsets were then fed into four different classifiers: support vector machine, random forest, logistic model tree and hidden naive bayes to classify melanoma in a test set of 102 images, which resulted in an accuracy of 88.24% and ROC area of 0.918. Comparative study carried out in this paper shows that the proposed feature extraction method outperforms three other wavelet-based approaches.","PeriodicalId":246460,"journal":{"name":"2010 International Conference on Digital Image Computing: Techniques and Applications","volume":"55 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2010-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126144151","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In this paper, a new technique is presented to enhance the blurred images obtained from Fluoresce in Angiography (FA) of the retina. One of the main steps in inspecting the eye (especially the deeper image of retina) is to look into the eye using a slit-lamp apparatus that shines a monochromatic light on to the retinal surface and captures the reflection in the camera as the retinal image. When further probing is required, such imaging is preceded by injecting a specialized dye in the eye blood vessels. This dye shines out more prominently in the imaging system and reveals the temporal as well as special behavior of the blood vessels, which, in turn, is useful in the diagnosis process. While most of the cases, the image produced is quite clean and easily used by the ophthalmologists, there are still many cases in which these images come out to be very blurred due to the disease in the eye such as cataract etc… in such cases, having an enhanced image can enable the doctors to start the appropriate treatment for the underlying disease. The proposed technique utilizes the Blind Deconvolution approach using Maximum Likelihood Estimation approach. Further post-processing steps have been proposed as well to locate the Macula in the image which is the zero-center of the image formed on the retina. The post-processing steps include thresholding, Region Growing, and morphological operations.
{"title":"Blind Restoration of Fluorescein Angiography Images","authors":"U. Qidwai, U. Qidwai","doi":"10.1109/DICTA.2010.35","DOIUrl":"https://doi.org/10.1109/DICTA.2010.35","url":null,"abstract":"In this paper, a new technique is presented to enhance the blurred images obtained from Fluoresce in Angiography (FA) of the retina. One of the main steps in inspecting the eye (especially the deeper image of retina) is to look into the eye using a slit-lamp apparatus that shines a monochromatic light on to the retinal surface and captures the reflection in the camera as the retinal image. When further probing is required, such imaging is preceded by injecting a specialized dye in the eye blood vessels. This dye shines out more prominently in the imaging system and reveals the temporal as well as special behavior of the blood vessels, which, in turn, is useful in the diagnosis process. While most of the cases, the image produced is quite clean and easily used by the ophthalmologists, there are still many cases in which these images come out to be very blurred due to the disease in the eye such as cataract etc… in such cases, having an enhanced image can enable the doctors to start the appropriate treatment for the underlying disease. The proposed technique utilizes the Blind Deconvolution approach using Maximum Likelihood Estimation approach. Further post-processing steps have been proposed as well to locate the Macula in the image which is the zero-center of the image formed on the retina. The post-processing steps include thresholding, Region Growing, and morphological operations.","PeriodicalId":246460,"journal":{"name":"2010 International Conference on Digital Image Computing: Techniques and Applications","volume":"4 6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2010-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126630330","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In this paper, to enable a fast and robust system for automatically recognizing license plates with various appearances, new and simple but efficient algorithms are developed to segment characters from extracted license plate images. Our goal is to segment characters properly from a license plate image region. Different from existing methods for segmenting degraded machine-printed characters, our algorithms are based on very weak assumptions and use no prior knowledge about the format of the plates, in order for them to be applicable to wider applications. Experimental results demonstrate promising efficiency and flexibility of the proposed scheme.
{"title":"Segmenting Characters from License Plate Images with Little Prior Knowledge","authors":"W. Jia, Xiangjian He, Qiang Wu","doi":"10.1109/DICTA.2010.48","DOIUrl":"https://doi.org/10.1109/DICTA.2010.48","url":null,"abstract":"In this paper, to enable a fast and robust system for automatically recognizing license plates with various appearances, new and simple but efficient algorithms are developed to segment characters from extracted license plate images. Our goal is to segment characters properly from a license plate image region. Different from existing methods for segmenting degraded machine-printed characters, our algorithms are based on very weak assumptions and use no prior knowledge about the format of the plates, in order for them to be applicable to wider applications. Experimental results demonstrate promising efficiency and flexibility of the proposed scheme.","PeriodicalId":246460,"journal":{"name":"2010 International Conference on Digital Image Computing: Techniques and Applications","volume":"184 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2010-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122072999","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Ubiquitous projection is the recent effort that tries to close the gap between the physical world and the virtual world by using a mobile projector. Using a projector and camera together has a conflict of preferred light conditions, making it difficult to implement a robust visual detector while retaining ubiquity. In this paper, we focus on techniques of visual object detection for a portable projector-camera system. The goal is to create a visual detector that requires no guidance from a user and is robust to different light conditions. Our investigation involves the multiscale concept using Canny edge detection as a representative detector. Five image simplification filters applied to the multiscale detection are examined for both accuracy and speed. In addition, preprocessing using histogram equalization and post processing are applied to ensure robustness in a real-world scenario, and to guarantee that the detection will always successfully detect objects using a constant set of parameters defined offline. Finally, we showed that using multiscale detection in a parallel manner can speed up the detection while not affecting the accuracy of the detection.
{"title":"Multiscale Visual Object Detection for Unsupervised Ubiquitous Projection Based on a Portable Projector-Camera System","authors":"Thitirat Siriborvornratanakul, Masanori Sugimoto","doi":"10.1109/DICTA.2010.109","DOIUrl":"https://doi.org/10.1109/DICTA.2010.109","url":null,"abstract":"Ubiquitous projection is the recent effort that tries to close the gap between the physical world and the virtual world by using a mobile projector. Using a projector and camera together has a conflict of preferred light conditions, making it difficult to implement a robust visual detector while retaining ubiquity. In this paper, we focus on techniques of visual object detection for a portable projector-camera system. The goal is to create a visual detector that requires no guidance from a user and is robust to different light conditions. Our investigation involves the multiscale concept using Canny edge detection as a representative detector. Five image simplification filters applied to the multiscale detection are examined for both accuracy and speed. In addition, preprocessing using histogram equalization and post processing are applied to ensure robustness in a real-world scenario, and to guarantee that the detection will always successfully detect objects using a constant set of parameters defined offline. Finally, we showed that using multiscale detection in a parallel manner can speed up the detection while not affecting the accuracy of the detection.","PeriodicalId":246460,"journal":{"name":"2010 International Conference on Digital Image Computing: Techniques and Applications","volume":"42 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2010-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129581374","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：