Tian-Hu Yu

Nonlinear unsharp masking methods for image contrast enhancement

In the unsharp masking approach for image enhancement, a fraction of the highpass filtered version of the image is added to the original image to form the enhanced version. The method is simple, but it suffers from two serious drawbacks. First, it enhances the contrast in the darker areas perceptually much more strongly than that in the lighter areas. Second, it enhances the noise and/or digitization effects, particularly in the darker regions, resulting in visually less pleasing enhanced images. In general, noise can be suppressed with lowpass filters, which are associated with the blurring of the edges. On the other hand, contrast can be enhanced with highpass filters, which are associated with noise amplification. A reasonable solution, therefore, is to use suitable nonlinear filters which combine the features of both highpass and lowpass filters. This paper outlines several new methods of unsharp masking based on the use of such nonlinear filters. Computer simulations have verified the superior results obtained using these filters. In addition, a new measure of contrast enhancement is introduced which quantitatively supports the improvement obtained using the proposed methods.

A new class of nonlinear filters for image enhancement

Two types of very simple two-dimensional nonlinear filters are introduced and applied to image contrast enhancement. The first type is based on a generalization of the Teagers algorithm. A theoretical analysis has shown that this type of nonlinear filter works like a local-mean-weighted highpass filter. Based on this analysis, a second type of nonlinear filter has been developed which works like local-mean-weighted bandpass filter. The proposed image contrast enhancement technique is based on combining the original image with its filtered version obtained using one of the two nonlinear filters. Very high quality enhancement has been achieved for natural images.<<ETX>>

Design of linear phase FIR notch filters

This paper investigates some approaches for designing one-dimensional linear phase finite-duration impulse-responses (FIR) notch filters, which are based on the modification of several established design techniques of linear phase FIR band-selective filters. Based on extensive design examples and theoretical analysis, formulae have been developed for estimating the length of a linear phase FIR notch filter meeting the given specifications. In addition, the design of two-dimensional linear phase FIR notch filters is briefly considered. Illustrative examples are included.

Transform amplitude sharpening: a new method of image enhancement

A new frequency-domain based technique is proposed. In the proposed method, the magnitude of the Fourier transform of the image is modified using a noval amplitude change function with the phase kept invariant. The inverse transform of the new 2D transform results in an enhanced image. The amplitude change function used is the Tukeys twicing function ( Exploratory Data Analysis , Addison-Wesley, Reading, MA, 1977). In many applications, multiple use of the twicing function is shown to yield better results. The proposed method is somewhat similar to the alpha-rooting method but does not suffer from certain objectionable artifacts associated with the latter and also exhibits less degradation due to noise. A postprocessing of the enhanced image generated using the proposed transform amplitude method via histogram modification has been found to improve the picture quality. A number of image enhancement examples are included illustrating the effectiveness of the proposed method.

A new nonlinear algorithm for the removal of impulse noise from highly corrupted images

Images are often corrupted by impulse noise from a noisy sensor or channel transmission errors. The goal of impulse noise removal is to suppress the noise while preserving the edges and the details. To this end, nonlinear techniques have been found to provide more satisfactory results in comparison to linear methods. The two often used nonlinear methods are the median filter and the outlier method. In median filtering the central pixel inside a window is replaced by the median value of all pixels in the window. However, median filtering not only removes impulse noise but also introduces distortion, in particular, for highly corrupted images. The outlier method is based on a detection-estimation strategy in which the impulse noise is detected first and then an average of all pixel values inside a window centered at the defected corrupted pixel is used as an estimate of its true value. This approach is also not effective for highly corrupted images. Recently, we proposed an efficient method to remove impulse noise from a highly corrupted image also based on a detection-estimation strategy, in which the impulse noise is detected first using a simple quadratic filter, and a selectively chosen local mean is used to estimate the true value of the corrupted pixel. A major limitation of this approach is that often it has to be applied twice to arrive at a satisfactory result. This second run not only increases computational cost but also reduce the quality of the processed image. In this paper, we analyze the limitation of this newly proposed method of impulse noise removal, and then advance a modified version which does not require a second run.<<ETX>>

A novel DPCM algorithm using a nonlinear operator

Differential pulse-code modulation (DPCM) has found many applications in both intraframe and interframe coding of images. To improve its performance, several adaptive techniques have been introduced, but they make the resulting DPCM systems very complicated. In this work, we introduce a nonlinear scheme of predictive coding to exploiting spatial redundancy and amplitude redundancy of images based on the characteristics of the human visual system. The proposed scheme consists of a novel nonlinear algorithm to generate new information and a new compandor model of uniform quantization.<<ETX>>

A new two-dimensional window

A two-dimensional (2-D) window is shown to be generated simply by applying the McClellan transformation to a one-dimensional (1-D) window, The frequency response of the transformed window is nearly circularly symmetric and preserves the features of its 1-D prototype.

Efficient detail-preserving method of impulse noise removal from highly corrupted images

Images are often corrupted by impulse noise due to faulty image acquisition devices or during transmission. The goal of impulse noise removal is to suppress the noise while preserving edges and details. In this paper, an efficient method is proposed to remove impulse noise from a highly corrupted image while preserving the detail features. This new method is based on a detection-estimation strategy, in which the impulse noise is detected first using a quadratic filter, and a selectively chosen local mean is used to estimate the true value of the corrupted pixel. Simulation results indicate that the new method is much more efficient and effective than the median filtering and the out-range methods.

Wavelet based hybrid image coding scheme

This paper describes a computationally efficient hybrid image coding scheme. The proposed hybrid coding scheme combines the advantages of simple hardware complexity of the block discrete cosine transform coding and the high performance of the embedded zerotree discrete wavelet transform coding. It is based on the following five key features: (1) one-scale discrete wavelet transform of the input image resulting in four subband images (LL, LH, HL and HH), (2) block discrete cosine transform coding of the LL band image, (3) nonlinear transformations of the residual image of the LL band, the LH, HL and HH images, (4) generation of zerotree structure of the nonlinearly transformed coefficients, and (5) lossless data compression via adaptive arithmetic coding of the zerotree structured data.

Stack X-tree image coding

This paper proposes a new image coding scheme, called the stack X-tree coding. It replaces the zero run in the stack-run coding by the X-tree. An X-tree is defined as a spatial hierarchical tree without significant descendants. By taking advantage of the inter-band dependencies, the X-tree can express more efficiently insignificant coefficients in a wavelet image than the zero run. Experimental results have shown that the performance of the proposed algorithm is better than that of the stack-run, and is highly competitive with that of SPIHT.