Article Preview
Top1. Introduction
The techniques of image processing stem from two principal applications, namely, improvement of pictorial information for human interpretation and processing of image data for storage, transmission, and representation for autonomous machine perception. Image enhancement, one of the most important image processing techniques, can be viewed as transforming one image to another to improve the interpretability of information for viewers, or to produce an enhanced image that is more suitable than the original for a specific application. It emphasizes or sharpens image features such as edges, boundaries, or increases contrast to make a graphic display more helpful for display and analysis (Gonzalez, Woods, & Eddins, 2011). Image Enhancement (IE) is a vital tool for researchers in the wide fields of research including (but not limited to) medical imaging, art studies, forensic, atmospheric sciences and vision systems. Generally, it is application-specific as IE techniques suitable for one problem might be inadequate for another. According to (Jain, 1989), the image enhancement technique can mainly be divided into four categories: algebra, spatial, transformation and pseudo-color processing. In this paper, image enhancement is done on the basis of spatial processing. Spatial operation deals with image in the spatial domain, based on direct manipulation of pixels in the image, divided into point processing and spatial filtering or mask processing. Point processing techniques include contrast stretching window slicing, and histogram modeling. This processing operates only on the intensity value at a point or pixel, not on a neighborhood of points. These methods have the disadvantage of treating the image in global fashion by not providing the details over a small area, while in many cases it is required to adopt the transformation to the local feature within different areas of the image. One advantage is that some of the point operations, such as histogram equalization (HE) and linear contrast stretching (LCS), are automatic methods. HE enhances the contrast of images by transforming the intensity values of the image so that the histogram of the improved image approximately matches the histogram of ideal image. HE is one of the most widely used simple methods for consumer electronics. Since it is indiscriminate in nature, several modifications have been done. These modifications are basically based on a technique of dividing the image of two or more parts with some specific method and equalize each part individually. However, enhancement of dark and low contrast, less and more details of images in a controlled way is still a challenge (Dhal & Das, 2017a). Contrast stretching is a process that expands the range of intensity levels in an image to make full use of possible values. Spatial filtering can be divided into image smoothing and image sharpening. Image smoothing such as average filtering and median filtering are generally used to eliminate image noise, but they make the edges hazy. The image sharpening is used to highlight the edges to facilitate object recognition. The well-known method of image sharpening is Unsharp Masking (UM) which enhances the image and detail information of image by emphasizing its high frequency contents (Gonzalez et al., 2011). Further, the cubic UM and adaptive UM have also been proposed and widely used (Ramponi, 1998).
A major problem with image enhancement is that a human interpreter is needed to judge whether an image is suitable for specific application or not. There is no specific benchmark against which the measurement of enhancement could be drawn. Automatic enhancement has some advantages like it provides local and global enhancement, it doesn’t employ any kind of user interaction, uses little or no external parameters which are sometimes difficult to fine tune, and it uses an objective evaluation criterion that produces an objective quality score (Munteanu & Rosa, 2004; Polesel, Ramponi, & Mathews, 2000).