ORTHOGONAL WAVELET DECOMPOSITION WITH MULTI-LEVEL THRESHOLDING FOR IMAGE ENHANCEMENT.

G.Kalpana 1 and G.Karuna 2 . 1. Assistant Professor, Department of MCA, CBIT, Hyderabad. 2. Associate Professor, Department of CSE, GRIET, Hyderabad. ...................................................................................................................... Manuscript Info Abstract ......................... ........................................................................ Manuscript History

The image enhancement techniques have been widely used to improve the quality of images with respect to human visual appearance, contrast and provide the better transform representation for future automated image processing. Image segmentation is the process of partitioning a 2D digital image into multiple segments. Different image segmentation techniques are available and they are mainly focused on representation of an image into more significant and easier to analyze. The quality of segmentation results are evaluated using PSNR and MSE values. Histogram equalization is used for image enhancement that improves the contrast in an image, in order to stretch out the intensity range. This method is useful in images with backgrounds and foregrounds that are both bright or both dark. However Histogram Equalization suffers from poor contrast and with noise. In the present work, a novel Multi-Level Threshold based algorithm for image segmentation is proposed. The main objective of proposed method is to reduce uneven contrast, unwanted noise, and unnecessary effects.
Image enhancement is a process that improves the definition of a video image by a program written in computer, so that it reduces the lowest gray-values in to black and highest gray-values into white. Mostly the images found from satellites and from traditional digital cameras lack in contrast and brightness due to the limited conditions during capturing of images. There are different types of noise obtained from the images. In image enhancement, the goal is to emphasize certain image features for subsequent analysis or for image display. Examples include contrast and edge enhancement, pseudo-colouring, noise filtering, sharpening and magnifying. Image enhancement technique is valuable in image analysis, extraction of image features, and displaying image. The enhancement process itself does not increase the inherent information content in the data. It simply emphasizes certain specified image characteristics. Enhancement algorithms are generally interactive intensive and application dependent. Some of the enhancement techniques are Contrast Stretching, Noise Filtering and Edge Enhancement. An image enhancement technique that works well for X-ray topographic images may not work well for MR images [1]. The enhancement methods can broadly be divided in to the following two categories [2]. Spatial Domain Methods and Frequency 1968 Domain Methods. In spatial domain techniques [3], directly deal with the image pixels. The wavelet transform [4,5] has a long and successful history as an efficient image processing method. Mixing problem can be improved with the directional extension for wavelets proposed in [6].
Contrast Stretching:-Some images (e.g. snow, clouds, forests, deserts, other hazy conditions of heterogeneous regions) are homogenous i.e., they do not have much change in their levels. In terms oSf histogram representation, they are characterized as the occurrence of very narrow peaks. This similarity is because of the scene incorrect illumination.
Ultimately the images hence obtained are not easily interpretable due to poor human perceptibility. This is because there exists only a narrow range of gray-levels in the image having provision for wider range of gray-levels. The contrast stretching methods [8] are designed exclusively for frequently encountered situations. Different stretching have been developed to stretch the narrow range to the whole of the available dynamic range.

Fig 1:-Contrast stretching
Noise Filtering:-Noise filtering is mainly advantageous to filter the excessive and unwanted information from an image. It is used to remove different kinds of noises also from the images. Mostly this feature is interactive. The filters such as high pass, low pass, mean, median are widely used.

Fig 2:-Noise Removal
Edge Enhancement:-Edge enhancement is one of the image processing filter [9, 10] which enhances the edge contrast of an image in an attempt to improve its seeming sharpness. The filter works by identifying sharp edge boundaries in the image, such as the edge between a subject and a background of a contrasting color, and increasing the image contrast in the area immediately around the edge.  [7] is a method in image processing of contrast adjustment using the image's histogram.

Fig 4:-Histogram Equalization
In general this technique improves the global contrast of most of the images, particularly when the usable data of the image is depicted by its close contrast values. The intensities can be distributed in good manner using this adjustment on the histogram. This permits the image areas of lower local contrast to obtain higher contrast. Histogram equalization does this by efficiently distributing the most frequent occurred intensity values. The method is helpful in images with backgrounds and foregrounds that are both bright and dark. Meticulously, this method can achieve the best views of x-ray images of bones structure. One of the beneficial from this method is that it is a reasonably straightforward process and an invertible operator. So in theory, if the histogram equalization function is known, then the original histogram can be recovered. The calculation is not computationally intensive.
There are considerable limitations in Histogram equalization. This method is indiscriminate. This method may increase the background noise contrast and decreasing the usable signal. In scientific imaging where spatial correlation is more important than intensity of signal (such as separating DNA fragments of quantized length), the small signal to noise ratio usually hampers visual detection. Histogram equalization is useful for scientific images such as satellite, thermal, x-ray images but it produces unexpected effects in most of the photographs. Histogram equalization can produce unwanted effects (i.e. visible image gradient) when applied to images with low color depth. This technique is is well suited for the images having brighter colors than palette size such as 6-bit gray-scale images.

Proposed Method:-
The proposed method can take advantage of all existing enhancement techniques for plain images and achieve excellent performance with respect to the removal of unwanted noise and to reduce unnecessary contrast. A Novel Multilevel thresholding Image enhancement technique has been proposed which is mainly based on orthogonal wavelet decomposition. Frequency based domain image enhancement is mainly focused to designate the exploration of mathematical functions in order to frequency and operate directly on the transform coefficients of the image, like discrete wavelet transform.

Structure Of Proposed
The Discrete Wavelet Transform [11, 12] is used the low pass and high pass filter to find the desire information. The most important information in the signal appears in high amplitudes and the least important information in the signal appears in very low amplitudes. Appropriate high pass and low pass filters are applied to the data at each level and also there is down sampling done at each level. The discrete wavelet transformation initially subdivided the image into four main areas [13]. The filtered data depends on the level of DWT where LL1 (Low-Low) sub band is depicted as an approximation data area and other sub bands are depicted as more detail information. The present sub bands are as follow: high-low (HL1), low high (LH1) and the high-high (HH1).  Orthogonal Wavelet Decomposition To Generate The Approximation Images:-Orthogonal wavelets used for discrete wavelet transform development. The DWT returns a data vector of the equal length as the input is. Generally, even in this vector many data are almost zero. i.e. the DWT decomposes the image into a set of wavelets or functions and they are orthogonal to its translations and scaling. Therefore we decompose such a signal to a same or lower number of the wavelet coefficient spectrum as is the number of signal data points. And this wavelet spectrum is appropriate for signal processing and compression.
The way the decomposition is shown below. Started with the original signal and fit the mother wavelet to it at the smallest scale. This produces what is called the first wavelet "detail" and a remainder called the "approximation" [14]. Then double the timescale of the mother wavelet (called dilation) and fit that to the first approximation. And so the process continues until the mother wavelet has been dilated to such an extent that it covers the entire range of the signal. The result is a set of details, each corresponding to an average timescale that doubles at each level, and an approximation. These details have an exceptional property called orthogonality and it means that they are completely independent of each other and therefore can be added together in any sequence. Furthermore, the variances can be added together to obtain the variance of the original signal.
Apply The Multi-Level Image Segementation Technique:-Finally a novel Multi-level thresholding technique is applied. In this technique, a thresholding value is taken as a reference and the pixel values are compared to this thresholding value. In general for a global thresholding technique, a single threshold value is taken. The pixel values less than the threshold value are shown as black and the pixel values more than threshold value are shown as white. In multi-thresholding technique, many values can be taken as reference. Here, two values are taken as reference and the pixel values are compared with these two values.
Multilevel thresholding is a method which divides a gray level image into many small regions. This technique finds more than one threshold for the given input image and segments the image into some brightness regions, and it corresponds to one background and several objects. This technique is well suited for color images having complex backgrounds.
The below Fig 4(b) shows the grayscale image and histogram equalized image and their corresponding histograms. An RGB image is converted into grayscale image in order to reduce the complexity of the technique used in the project. To find the histogram of the image, "imhist" command is used. In order to find the histogram equalized image, "histeq" command used. And found that the histogram of the equalized image is stretched when compared to the grayscale image, stating that the image contrast has increased. Fig 4(c) shows the wavelet decomposition technique. In this technique, the image is being divided into four. They are: Approximation image, Horizontal image, Vertical image, Diagonal image. 1972

Conclusion:-
As an initial attempt, here is the motivation and solution for the problem of poor contrast, noise in the image. Initially a basic scheme is given and followed by existing Histogram Equalization applied. However it is very inefficient to achieve image enhancement and also fails to removal noise and reduce contrast. Wavelets have their energy concentrated in time and are well suited for the analysis of transient, time-varying signals. The wavelet transformations are well suited because most of the real life signals come upon time varying in nature. Further investigated enhancement techniques have been applied. The Orthogonal Wavelet Decomposition used first and combined this with multi-level thresholding technique. The thorough analysis indicated that, the proposed system is successful in achieving image enhancement than the existing methods in terms of removal of unnecessary noise and reduced unwanted contrast and Extensive experimental results demonstrate the efficiency of our solution.