Efficient text segmentation and adaptive color error diffusion for text enhancement

Multiscale error diffusion is superior to conventional error diffusion methods in digital halftoning as it can eliminate directional hysteresis completely. However, there is a bias to favor a particular type of dots in the course of the halftoning process. A new multiscale error diffusion method is proposed to improve the diffusion performance by reducing the aforementioned bias. The proposed method can eliminate the pattern noise in flat regions and the boundary effect found in some other conventional multiscale error diffusion methods. At the same time, it can preserve the local features of the input image in the output. This is critical to quality, especially when the resolution of the output is limited by the physical constraints of the display unit.

Color halftoning is a method for representing continuous tone color images with limited colors. To implement halftone, error diffusion is one of the most popular algorithms due to its highperformance results. In the conventional error diffusion, the fixed thresholds and filter weights are used. In the process, many artifacts are introduced in the halftone images as side effects. With the color aspect, each of the three color planes is independently calculated during the process of traditional error diffusion. However, the pattern of dots formed independently by combining the dots of each of color planes is often not visually pleasing because no effort is made to insure that the dots of each of the different color planes are ideally distributed relative to the dots of the other color planes. In this paper, we present an integrated error diffusion algorithm in which variable filter weights, threshold matrix, serpentine scan path and multi-color halftoning are used to reduce the noises generated during the halftone process. Our experimental results showed that worm-like artifacts, structural artifacts and start-up artifacts are successfully decreased and the color halftoned images are more comfortable to the human visual system.

A method has been developed to induce symmetric edge enhancement in error diffusion with threshold modulation. The existing threshold modulation algorithm induces edge enhancement by subtracting a term proportional to the input image from the threshold. The edge enhancement that results in asymmetric due to the asymmetric shape of the error diffusion filter. The new method induces a symmetric edge enhancement by subtracting a term from the threshold that is proportional to a filtered version of the input image. The filtering is accomplished with a recursive filter, applied during the error diffusion algorithm, which cancels out the error diffusion filter and imposes a symmetric sharpening filter. The result is a symmetric edge enhancement that is more pleasing to the eye.

Taking colorimetric color reproduction into account, the conventional error diffusion method is modified for color digital halftoning. Assuming that the input to a bilevel color printer is given in CIE-XYZ tristimulus values or CIE-LAB values instead of the more conventional RGB or YMC values, two modified versions based on vector operation in (1) the XYZ color space and (2) the LAB color space were tested. Experimental results show that the modified methods, especially the method using the LAB color space, resulted in better color reproduction performance than the conventional methods. Spatial artifacts that appear in the modified methods are presented and analyzed. It is also shown that the modified method (2) with a thresholding technique achieves a good spatial image quality.

Digital halftoning is essential to print or display a continuous tone with binary devices. Error Diffusion (ED) has been a major stream to produce a pictorial image on inkjet prints, but is not always acceptable for commercial printing or EP printer, because it generates the dispersed dots hard to print.This paper proposes a novel method for generating ordered dither mask using “magic square”. Magic square provides spatially balanced threshold values in row, column, and diagonal directions by the rule of “even-sum”. The proposed magic square satisfies the energy minimum criterion in the local area and is expandable to a large scale for sufficient gray scales. We selected the magic square candidates with Fourier spectra concentrated in the middle spatial frequency to produce a smoothed gradation suitable for LBP or commercial printing. The paper introduces a systematic method for expanding small mask to large mask by “master-slave” algorithm. The experimental results show the improved grayscale reproduction with reduced artifacts in comparison with the conventional Bayer dither or ED.

Currently, many low-cost computers can only simultaneously display a palette of 256 colors. However, this palette is usually selectable from a very large gamut of available colors. For many applications, this limited palette size imposes a significant constraint on the achievable image quality. We propose a method for designing an optimized universal color palette for use with halftoning methods such as error diffusion. The advantage of a universal color palette is that it is fixed and therefore allows multiple images to be displayed simultaneously. To design the palette, we employ a new vector quantization method known as sequential scalar quantization (SSQ) to allocate the colors in a visually uniform color space. The SSQ method achieves near-optimal allocation, but may be efficiently implemented using a series of lookup tables. When used with error diffusion, SSQ adds little computational overhead and may be used to minimize the visual error in an opponent color coordinate system. We compare the performance of the optimized algorithm to standard error diffusion by evaluating a visually weighted mean-squared-error measure. Our metric is based on the color difference in CIE L *a *b , but also accounts for the lowpass characteristic of human contrast sensitivity.

In Digital Subtraction Angiography (DSA) image registration algorithm, the precision of the control points as well as their number and the distribution in image determine the accuracy of geometric correction and registration. Control points usually adopt the grid points; however, a more effective method is to extract control points adaptively according to the image feature. In this paper, a control point’s selection algorithm of DSA images is proposed based on adaptive multi-Scale vascular enhancement, error diffusion and means shift algorithms. Experimental results show that the proposed algorithm can adaptively put the control points to blood vessels and other key image characteristics, and can optimize the number of control points according to practical needs, which will ensure the accuracy of DSA image registration.

In this paper, we propose a text extraction method from camera-captured document style images and propose a text segmentation method based on a color clustering method. The proposed extraction method detects text regions from the images using two low-level image features and verifies the regions through a high-level text stroke feature. The two level features are combined hierarchically. The low-level features are intensity variation and color variance. And, we use text strokes as a high-level feature using multi-resolution wavelet transforms on local image areas. The stroke feature vector is an input to a SVM (support vector machine) for verification, when needed. The proposed text segmentation method uses color clustering to the extracted text regions. We improved K-means clustering method and it selects K and initial seed values automatically. We tested the proposed methods with various document style images captured by three different cameras. We confirmed that the extraction rates are good enough to be used in real-life applications.

Textual information contained in images is a valuable source of high-level semantics for image indexing and retrieval. This paper proposes a new method to detect and segment text from complex images. First, a density-based clustering method is employed to discover the candidate text regions. The clustering method is from data mining area. It computes the density distribution of overall image and makes spatial connective pixels with similar color/grayscale into one region. The clustered regions are deemed as candidate text regions. Then simple heuristics are applied to delete those obvious non-text regions from the candidate. But there still exits a few non-text regions in the candidate. Therefore a texture-based method is used to select text regions from the filtered candidate text regions. Considering the time complexity of density computation in clustering step, an approximate algorithm is designed to improve the efficiency. Experimental result shows the method is robust to variations in text font, orientation, language, and size.

A digital halftoning method is proposed to diffuse error with a more symmetric error distribution by making use of the concept of multiscale error diffusion. It can improve the diffusion performance by effectively reducing directional hysteresis. The diffusion is row-oriented father than frame-oriented and hence can reduce a lot of computation effort as compared with conventional multiscale error diffusion schemes. This makes it possible for real-time applications.

A digital halftoning method is proposed to diffuse error with a more symmetric error distribution by making use of the concept of multiscale error diffusion. The method can improve the diffusion performance by effectively reducing directional hysteresis. The diffusion is row-oriented rather than frame-oriented and hence can reduce the latency and computational effort as compared with conventional multiscale error diffusion schemes. This makes it possible for real-time applications.

Web has become increasingly oriented to multimedia content. Most information on the web is conveyed from images. Text localization in web image plays an important role in web image information extraction and retrieval. Current works on text localization in web images assume that text regions are in homogenous color and high contrast. Hence, the approaches may fail when text regions are in multi-color or imposed in complex background. In this paper, we propose a text extraction algorithm from web images based on the probabilistic candidate selection model. The model firstly segments text region candidates from input images using wavelet, Gaussian mixture model (GMM) and triangulation. The likelihood of a candidate region containing text is then learnt using a Bayesian probabilistic model from two features, namely, histogram of oriented gradient (HOG) and local binary pattern histogram Fourier feature (LBP-HF). Finally best candidate regions are integrated to form text regions. The algorithm is evaluated using 155 non-homogenous web images containing around 600 text regions. The results show that the proposed model is able to extract text regions from non-homogenous images effectively.

Treating color patterns in a geometric morphometrics framework reveals rapid rates of color evolution that are explained by a combination of intrinsic organismal features (color variation among patches) and geography within a cosmopolitan radiation of birds.

The objective of this work is to reconstruct high quality gray-level images from halftone images, or the inverse halftoning process. We develop high performance halftone reconstruction methods for several commonly used halftone techniques. For better reconstruction quality, image classification based on halftone techniques is placed before the reconstruction process so that the halftone reconstruction process can be fine tuned for each halftone technique. The classification is based on enhanced 1D correlation of halftone images and processed with a three- layer back propagation neural network. This classification method reached 100 percent accuracy with a limited set of images processed by dispersed-dot ordered dithering, clustered-dot ordered dithering, constrained average, and error diffusion methods in our experiments. For image reconstruction, we apply the least-mean-square adaptive filtering algorithm which intends to discover the optimal filter weights and the mask shapes. As a result, it yields very good reconstruction image quality. The error diffusion yields the best reconstructed quality among the halftone methods. In addition, the LMS method generates optimal image masks which are significantly different for each halftone method. These optimal masks can also be applied to more sophisticated reconstruction methods as the default filter masks.

In this paper a real time approach to reduce speckle noise and enhance structures in speckle corrupted ultrasound image is proposed. It utilizes nonlinear anisotropic diffusion for speckle removal and edge enhancement. Unlike the conventional anisotropic diffusion performed in four different directions, here diffusion is performed in eight different directions. The algorithm has been proficiently implemented using a low cost Texas Instruments (TI) Digital Signal Processor (DSP) of series TMS320C64x+. The approach speeds up the computation by utilizing the ability of Very Long Instruction Word (VLIW) processor to perform multiple operations in parallel. In addition, a new feature to select a Speckle Reduction Imaging (SRI) mode is introduced which facilitates different degree of smoothing depending on the requirement. The experimental results show that with an speckled input image of size 640 × 480 the algorithm is capable of supporting 105 frames per second (fps). This method also promises better image quality than window based filtering techniques [1]–[4].