Background appearance of images for human scenes 2)

Background
Overview            Image processing the global area
that involve 1) enhance and improving the visual appearance of images for human
scenes 2) preparing images for measurement measuring features and current
structures. When atmospheric moisture effect on the scenes it will safely
degrades the visibility of outdoor scenes it is called haze1. Haze fetches trouble to many computer vision and affect
frequently on graphics applications as it minify the clarity of the scene2. So attenuation (decreases the disparity) and the air light (increases
the whiteness) are the two fundamental phenomena those cause a haze, figure 1 illustrate both attenuation and air light
phenomena.Figure
1 Attenuation and air light phenomenaThe
collaborations between atmosphere and the light causes fog and haze like
absorption, dispersion, and emission, but basically they are different in the
sizes and types of scattering particles3.In
recent year there are many techniques used in many computer vision applications
that recover the color and contrast of the scene to remove this haze, such as
outdoor surveillance, object detection, consumer electronics, enhancement etc.1.Usually when remove the haze, which is called
dehazing, is commonly performed under the physical degradation model, which
presuppose a solution of problem is not reversed4.

Most
picture dehazing algorithms consider utilizing a hard brim presumptions or
customer contribution to evaluate atmospheric light5. According to (C. Chengtao1,) 6 they are characterized the various dehazing picture approaches
into two general classifications i.e. picture improvement and regain physical
model. Image-dehazing methods can be roughly categorized into two kinds: the
methods based on computer visions and those based on physical models. The
advantage of computer-visionbased methods is that they can do the dehazing
process by utilizing only one single image7.Haze
Models

In
the field computer vision and image processing the using of haze creation model
it takes a broadly place. This model in most cases used for the development of
image in the existence of bad atmospheric situations. The particular size of particles
in the atmosphere is between 1-10 ?m. So the existence of these particles in
the aerosol effect on the quality of image.2

The deviations
that get into the atmosphere are also observed from the dreadful weather circumstance
consists of haze, fog, mist, a nice decomposition of smoke, or other media from
the outdoor landscape, due to this several problems occur, such as automated
oversight system, the outdoor identification system, distant sensing systems, and the smart conveyance system, such as traffic
observation systems and travelling vehicle data recorders are strongly affected8.

Haze
reduces the contrast and saturation degrades the quality of preview and
captured image. Haze attenuates the mild pondered from the scenes, and
similarly blends it with some additive light inside the atmosphere.

The
goal of haze elimination is to enhance the contemplated light (i.e., the scene
colors) from the mixed mild

 

Haze
in Digital Images

Digital
pictures caught in outdoor landscape condition are effectively contaminated by
haze, which will degrade the transferred information. Haze is a physical
phenomenon that darkens scenes, decreases vision, and changes colors.

 

2.2.1
Haze Definition

Haze
is constituted of aerosol which is a dispersed system of small particles
suspended in gas. Haze has a various set of sources including volcanic fiery
debris, foliage exudation, combustion products, and sea salt. The particles formed by these sources react rapidly to changes in
relative moisture and go about as cores (focuses) of tiny water beads when the
dampness is high. Haze particles are bigger than air atoms yet littler than
haze beads.

2.2.2
Mechanism of Atmospheric Scattering

 

 Figure 2.1: Scattering of light by atmospheric
particles 9(McCartney, 1975)

The
study of the react of illumination with the atmosphere (and hence climate) is
widely known as aerial optics. Aerial optics lies at the core of the most
magni?cent visual expertise known to man, including, Sunrise colors and sunset,
the blueness of the pure sky, and the rainbow10 11. (see Minnaert (1954) and Henderson(1977)).

The
major source of materials as background in this field is the works of McCartney
(1975) and Middleton (1952) which, despite its history, was an excellent review
of previous work.

The
main features of light, such as density and color, were altered through its
connections with the atmosphere. These interactions can be broadly classified
into three classes: dispersion, absorption, and emissions12.It therefore
behaves like appoint source of light. The exact scattering function is closely related
to the ratio of particle size to wavelength of incident light. (Adapted from Minnaert
(1954)). In
atmospheric scattering, the transmission properties of light can be categorized
into two mechanisms which are airlight and direct transmission: 2.2.2.1
Airlight  The
existence of particles in the aerosol those generated by the haze effect on the
quality of image. In this case, when the image is taken, the camera absorbs the
light close to and scattered by these atmospheric particles.  So the technique is called as airlight; which
is the first components of transmission properties. Suppose that this haze
demonstrate is straight model. From the linearity’s definition in this model
the change occurs just on pixel position.  2.2.2.2
Direct Transmission  In computer vision and image processing, the second
components of transmission properties of is the direct transmission of light
from the object surface that describes the beam light attenuation as it
traverses through the atmosphere from a scene point to the camera13.So haze
is the mix of the two fundamental phenomena direct attenuation and the air
light. So the formation of hazy image in12 13 14 is broadly written and it is describe as follows: I(x) = J(x)*t(x) + A*(1-t(x))                                                                                       (1)
 Where
I (x) attitude for the spotted density of the xth pixel, J (x) is the radiation
sight (the genuine color that we need to retrieval), A Is the light of the
universal atmosphere, and t Is the medium transmission that portrays the
section of the light that does not spread and reaches the camera.First
expression in the equation, J(x)*t(x) is called the direct attenuation; the
second expression, A*(1-t(x)) is called Airlight.In
vision systems, the transmission can be expressed as:t(x) = e??(x)d(x)                                                                                                   (2) Where
d(x) is the distance between the viewer and an object and ?(x) is the
scattering coef?cient which is dependent on turbidity T and wavelength ?. In
haze condition, the scattering coef?cient is generally assumed to be
independent of wavelength13. Thus, the coef?cient varies with turbidity T. Since t(x) (0 < t(x) ? 1) here does not correspond to the wavelength-depending physical atmospheric transmission, transmissions are the same for all RGB channels15.Particles in space At most weather cases vary in the kinds and sizes of the particles entangled and their focuses in space. Much effort has been made to measure particle sizes and concentrations for an assortment of conditions, so bigger particles create an assortment of climate conditions which illustrate more in Table 1. Given the little size of air particles, relative to the wavelength of obvious light, dissipating because of air is somewhat negligible12.Dehazing MethodsHaze can transform a colored picture into a white-and-ashy one, causing lost picture details and decrease in disparity. Likewise haze trouble numerous applications, including targeted direct monitoring and indirect confession, tracking, and measurements. Picture dehazing can take off haze from the pictures, increment the scene vision, and enhance the general visual impact16.The great challenge that rests with mathematical ambiguity is the removal of haze. Though dehazing images is very important in computer vision applications. Consequently, most of researcher strive to attitude these challenging tasks and suggested a variety of dehazing algorithm. Dehazing methods can be collected into two categories that are single image haze removal which required only single image as input and multiple image haze removal which are take multi images two, three, or more of the same sight. Both methods comes under many categories are described in the following diagram.Single Image Dehazing Haze removal algorithm which required only single image as input can classified single image into three major types: 1) Algorithms based on priors or hypotheses. This type of methods takes off fog from image during valuing parameters of the model fog imaging, which can fulfill satisfying outcomes. 2) Enhancement image on the basis of image processing, since these methods at most focus on picture enhancement and consider little of the imaging model of debased pictures, so when the scene is unpredictable unsatisfying outcomes will obtain. 3) Dehazing based fusion strategy. In their technique, two information sources got from the first picture authentic are weighted by three standardized weight maps (luminance, chromatic and saliency) and mingled in a multi-scale combination ?nally to take out haze impacts17.Recently the researchers are more interesting with this method. This method classifies techniques to the following categories.Dark Channel Prior (DCP) Different dehazing algorithms of single image dark channel have planned to handle the issue of picture right of passage in a quick and effective way. Such algorithms depend upon the dark channel earlier hypothesis towards the air light the estimation of  which offers itself as a urgent parameter towards dehazing.18.