| (7 intermediate revisions by 3 users not shown) | |||
| Line 1: | Line 1: | ||
| − | + | ===Image Matching=== | |
| − | + | Maximum likelihood estimates can be used in image matching (edge template matching and gray-level image matching). This can be applied to stereo matching and feature tracking. More about this topic can be found here ... [http://portal.acm.org/citation.cfm?id=568226] | |
| − | + | ||
| − | + | Maximum likelihood can also be used in image reconstruction or restoration. Surprisingly, I found the usage of this in compression artifact removal also. See this paper [http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?tp=&arnumber=901102&isnumber=19490] | |
| − | + | ===Face Recognition=== | |
| − | [ | + | Fisher Linear Discriminant (FLD) is widely used in face recognition. Here is a paper for reference: [http://ieeexplore.ieee.org/Xplore/login.jsp?url=/iel4/5726/15322/00711956.pdf]. Also variants of FLD are used for face recognition such as DiaFLD [http://linkinghub.elsevier.com/retrieve/pii/S0925231206000877]. It has been observed that FLD works better than Principal Component Analysis in classifying the facial features. |
| − | + | See also: [[Face detection vs Face recognition_OldKiwi]] | |
| − | + | ===Image Segmentation=== | |
| + | Image Segmentation is performed by conventional graphical methods, but many a times, some pixels not belonging to the same object are classified into the same segment. Also, in images where a wide background is separated by a thin boundary line, image segmentation can be performed by obtaining features from FLD. I experimented this personally and found that the results are better than the conventional methods. This paper gives a starting point in doing this [http://ict.ewi.tudelft.nl/~duin/papers/asci_99_diabolo.pdf]. | ||
| − | + | Here is an [[Image Segmentation Lab_OldKiwi|example]] of a clustering method used for image segmentation. | |
| + | |||
| + | ===Classification/Clustering in Remote Sensing=== | ||
| + | In remote sensing area, classification methos have been widely used for classifying multispectal or hyperspectal image data. The main purposes of classification using remotely-sensed image are to make thematic maps such as landuse and landcover etc. and to reduce the cost for surveying. We can found lots of information about this in Web, and here are some useful links. | ||
| + | |||
| + | [http://www.cac.science.ru.nl/research/publications/PDFs/tran2005.pdf] | ||
| + | [http://www.microimages.com/getstart/pdf/classify.pdf] | ||
| + | |||
| + | ===Content-based Image/video Retrieval=== | ||
With the increasing developments in memory technology, there has been a huge increase in the storage capacities of cameras, PDAs etc. This has led to a dramatic growth in personal image collections, and the image database sizes on the whole. The indexing of the databases however is not organized. So, retrieving images based on their names/indexes is very difficult. Hence, the images have to be retrieved by their content. | With the increasing developments in memory technology, there has been a huge increase in the storage capacities of cameras, PDAs etc. This has led to a dramatic growth in personal image collections, and the image database sizes on the whole. The indexing of the databases however is not organized. So, retrieving images based on their names/indexes is very difficult. Hence, the images have to be retrieved by their content. | ||
| − | Pattern recognition plays an important role in retrieving of images. The features like color, texture, brightness, segments, shapes etc. of the images are used for retrieval. The appropriate feature selection is imperative to achieve good accuracy. Some of well-known image retrieval systems are 'imgseek', 'blobworld' [http://citeseer.ist.psu.edu/72984.html]. | + | Pattern recognition plays an important role in retrieving of images. The features like color, texture, brightness, segments, shapes etc. of the images are used for retrieval. The appropriate feature selection is imperative to achieve good accuracy. Some of well-known image retrieval systems are 'imgseek', 'blobworld' [http://citeseer.ist.psu.edu/72984.html], 'Bag of features' [http://vision.ucla.edu/~vedaldi/code/bag/bag.html]. |
Pattern recognition can also be used to retrieve videos based on their content. This is more complex as compared to the image retrieval, because of the additional dimensions (time, motion). Several attempts have been made to develop good video retrieval systems. In some systems, classifying videos into categories like action, sports, animation, news/lectures etc. helps to reduce the computation complexity and the retrieval time, by retrieving the videos from the appropriate categories. | Pattern recognition can also be used to retrieve videos based on their content. This is more complex as compared to the image retrieval, because of the additional dimensions (time, motion). Several attempts have been made to develop good video retrieval systems. In some systems, classifying videos into categories like action, sports, animation, news/lectures etc. helps to reduce the computation complexity and the retrieval time, by retrieving the videos from the appropriate categories. | ||
Latest revision as of 13:40, 25 April 2008
Contents
Image Matching
Maximum likelihood estimates can be used in image matching (edge template matching and gray-level image matching). This can be applied to stereo matching and feature tracking. More about this topic can be found here ... [1]
Maximum likelihood can also be used in image reconstruction or restoration. Surprisingly, I found the usage of this in compression artifact removal also. See this paper [2]
Face Recognition
Fisher Linear Discriminant (FLD) is widely used in face recognition. Here is a paper for reference: [3]. Also variants of FLD are used for face recognition such as DiaFLD [4]. It has been observed that FLD works better than Principal Component Analysis in classifying the facial features.
See also: Face detection vs Face recognition_OldKiwi
Image Segmentation
Image Segmentation is performed by conventional graphical methods, but many a times, some pixels not belonging to the same object are classified into the same segment. Also, in images where a wide background is separated by a thin boundary line, image segmentation can be performed by obtaining features from FLD. I experimented this personally and found that the results are better than the conventional methods. This paper gives a starting point in doing this [5].
Here is an example of a clustering method used for image segmentation.
Classification/Clustering in Remote Sensing
In remote sensing area, classification methos have been widely used for classifying multispectal or hyperspectal image data. The main purposes of classification using remotely-sensed image are to make thematic maps such as landuse and landcover etc. and to reduce the cost for surveying. We can found lots of information about this in Web, and here are some useful links.
Content-based Image/video Retrieval
With the increasing developments in memory technology, there has been a huge increase in the storage capacities of cameras, PDAs etc. This has led to a dramatic growth in personal image collections, and the image database sizes on the whole. The indexing of the databases however is not organized. So, retrieving images based on their names/indexes is very difficult. Hence, the images have to be retrieved by their content.
Pattern recognition plays an important role in retrieving of images. The features like color, texture, brightness, segments, shapes etc. of the images are used for retrieval. The appropriate feature selection is imperative to achieve good accuracy. Some of well-known image retrieval systems are 'imgseek', 'blobworld' [8], 'Bag of features' [9].
Pattern recognition can also be used to retrieve videos based on their content. This is more complex as compared to the image retrieval, because of the additional dimensions (time, motion). Several attempts have been made to develop good video retrieval systems. In some systems, classifying videos into categories like action, sports, animation, news/lectures etc. helps to reduce the computation complexity and the retrieval time, by retrieving the videos from the appropriate categories.
