Segmentation of cursive handwriting is one of the most challenging problems in the area of handwritten character recognition. In this paper, we propose a novel approach towards character segmentation in a handwritten ...
详细信息
Segmentation of cursive handwriting is one of the most challenging problems in the area of handwritten character recognition. In this paper, we propose a novel approach towards character segmentation in a handwritten document. It is based on the vertex characterization of outer isothetic polygonal covers so that each cover corresponds to a particular word or part of a word. The proposed method has the potential to segment skewed text without deskewing them. Experiment is done on several Bangla handwritings of different individuals. The average success rate is 96.04\%. This method can be considered as a significant preprocessing step towards the development of a handwritten Bangla OCR system.
In this paper, we propose a fuzzy rule based method for extensive segmentation of documents, with the view of labeling various components like page header, title, figure and tables, column divider, text etc. The fuzzy...
详细信息
In this paper, we propose a fuzzy rule based method for extensive segmentation of documents, with the view of labeling various components like page header, title, figure and tables, column divider, text etc. The fuzzy rules are learned using evolutionary computation, based on labeled document images serving as ground truth. The proposed fuzzy rules exploit the placement heuristics among the various components in a technical paper. A genetic algorithm involving individual fitness functions of all components selects the best set of rules. The effectiveness of the proposed method is validated through a case study involving 300 documents. It is also shown that this approach can be extended to segmentation of magazine pages.
A fundamental problem in imageprocessing is image segmentation. The conventional image segmentation methods, to some extent, all suffe from the problem of inaccurate segmentation. A slightly easier and more approacha...
详细信息
A fundamental problem in imageprocessing is image segmentation. The conventional image segmentation methods, to some extent, all suffe from the problem of inaccurate segmentation. A slightly easier and more approachable problem - interactive segmentation - has also received a lot of attentions over the years. In this paper, we propose a fast and simple interactive image segmentation technique. This segmentation process is conducted in two modules. First, the original image is pre-segmented by a low level segmentation method. Second, some portion of the object-of-interest is marked. The proposed method identifies all the similar-colored objects present in the image. In this work, color feature is used to measure the closeness between two regions and accordingly the label of the unmarked region is decided.
This paper reports the development of a software suite to be accessed in future with any General Packet Radio Service (GPRS) enabled mobile phone or Personal Digital Assistant (PDA) for the extraction and analysis of ...
详细信息
We present a video compression scheme using epitome based texture coding that uses a low quality video as side information and improves it by using the epitome. The side information is sent at different levels of quan...
详细信息
We present a video compression scheme using epitome based texture coding that uses a low quality video as side information and improves it by using the epitome. The side information is sent at different levels of quantization and resolution to optimize the quality against the bit rate. The concept of motion threading is used for propagation of epitome information from one frame to another. The proposed scheme shows promising results for low bit rate side information case because of the reduction in the level of blurring due to the introduction of high frequency information derived from the epitome.
Segmentation of unconstrained handwritten word into different zones (upper middle and lower) and characters is more difficult than that of printed documents. This is mainly because of variability in inter-character di...
详细信息
Segmentation of unconstrained handwritten word into different zones (upper middle and lower) and characters is more difficult than that of printed documents. This is mainly because of variability in inter-character distance, skew, slant, size and curved like handwriting. Sometimes components of two consecutive characters may be touched or overlapped and this situation complicates the segmentation task greatly. In indian languages such touching or overlapping occurs frequently because of modified characters of upper-zone and lower-zone. In this paper we propose a simple method to segment unconstrained handwritten Bangla word. We achieved 81.41% success rate in the proposed system.
This paper proposes a method for personal identification based on iris recognition. The iris segmentation is obtained by using an integro-differential operation. The segmented iris is then normalised and actually a sm...
详细信息
This paper proposes a method for personal identification based on iris recognition. The iris segmentation is obtained by using an integro-differential operation. The segmented iris is then normalised and actually a small portion of the normalised portion is used for feature extraction. Three types of features are used: GLCM based features, Edge based features and Local Directional Pattern. We present a comparison of the performances of the above mentioned features. It is also shown experimentally that the half-way iris patterns exhibit a symmetry about the vertical axis. The multiclass recognition problem is reduced to a two class verification problem. Experimental results show that our proposed method has encouraging performance.
Digital in painting is the technique of filling in the missing regions of an image using information from the surrounding area in a visually indistinguishable way. In this paper, we try to improve the Exemplar based m...
详细信息
Digital in painting is the technique of filling in the missing regions of an image using information from the surrounding area in a visually indistinguishable way. In this paper, we try to improve the Exemplar based method by manipulating the values of various parameters like patch size, shape and size of the mask. We present an analysis of the impact of various geometric parameters on the quality of in painted images. image denoising refers to the removal of unwanted noise from the images. In most cases, the images which need to be in painted are noisy, which makes it necessary to eliminate noise and fill in the missing regions from neighboring pixels. Therefore, filling in of missing regions and removal of noise are the two very important topics in imageprocessing. This paper also addresses the issue of performing both in painting and denoising simultaneously using two different approaches: pipelined approach and interleaved approach. The effectiveness of these approaches is demonstrated with a number of results on various images.
We present a new concept for quantum computation and algorithms along with its theoretical understanding of quantum physics. We formulates the quantum logic and cognition of mind in general as the eigen value problems...
详细信息
We present a new concept for quantum computation and algorithms along with its theoretical understanding of quantum physics. We formulates the quantum logic and cognition of mind in general as the eigen value problems with the introduction of cognitive quantum number (CNQ), which interprets that the measurement act in the quantum domain parallels a decision process in domain of logic and it translates between the two. We also described the intelligent code of mind considering its operation as quantum mechanical in nature, that helps the modeling of information processing in case of quantum computing. Synthesis and spectral response of nano-particle towards quantum dot formation is also studied.
In this paper, a hybrid fingerprint matching algorithm based on phase correlation and minutiae points is proposed to improve the matching accuracy in fingerprint verification system. The phase correlation function whi...
详细信息
In this paper, a hybrid fingerprint matching algorithm based on phase correlation and minutiae points is proposed to improve the matching accuracy in fingerprint verification system. The phase correlation function which operates in the Fourier domain is used to determine the rotational offset while the core point detection algorithm is used to provide translational adjustment. Finally to match the aligned input and template minutiae sets, a Weighted Average Distance (WAD) measure is introduced. Experimental results show that the proposed method has an Equal Error Rate of 1.02% and 0.77% on two public domain databases FVC2002-DB1a and FVC2002-DB2a respectively.
暂无评论