FORM 2
THE PATENTS ACT 1970
(39 of 1970)
AND
The Patents Rules, 2003
COMPLETE SPECIFICATION
(See section 10 and rulel3)
1. TITLE OF THE INVENTION:
"METHOD AND SYSTEM FOR IMAGE ENHANCEMENT USING WIGNER DISTRIBUTION"
2. APPLICANT:
(a) NAME: KPIT CUMMINS INFOSYSTEMS LIMITED
(b)NATIONALITY: Indian Company incorporated under the Companies Act, 1956
(c) ADDRESS: 35 & 36 Rajiv Gandhi Infotech Park, Phase 1, MIDC, Hinjewadi, Pune - 411057, Maharashtra, India.
3. PREAMBLE TO THE DESCRIPTION:
The following specification describes the invention and (he manner in which it is to be performed.

Field of Invention:
The present invention generally relates to image processing and more particularly, to image enhancement. Further, the present invention provides for enhancing images, specifically low light condition images and night vision images, for human visual enhancement and inspection, and other such applications.
Background of Invention:
Image enhancement or the process of improving upon the quality of a digital image plays a significant role in interpretability or perception of information in images for human visualization and inspection or for providing a better input for various image processing techniques. Image enhancement is of prime importance in various fields like medical, astronomy, seismology, safety applications, etc,
The various existing systems, often utilize elementary methods for enhancement of images. These techniques often result in improper enhancement of the image, leading to a loss of data and hence, cumulative reduction in the accuracy with regards to further processing. The methods utilized are complex, require multiple image modalities, increase the hardware overhead and are not very efficient with noise reduction. Due to variations in weather conditions such as snow, fog, rain, low light, etc. images captured by the camera may contain high level of noise. Thus, an ideal image restoration system must consider various levels of noise and must provide a means for eliminating a wide range of noise and improved contrast, sharpness and brightness for better image processing and restoration.
Summary:
The present invention provides for a method and system for image enhancement using Wigner distribution. The enhancement technique of the invention provides for processing of images with regards to noise reduction and hence, providing a better clarity of images for human visual inspection and such other applications. The technique works well with different signal to noise ratios ranging from -1.58dB to 20dB and helps overcome poor image quality due to a variety of factors like fog, snow, rain, low light, etc. The technique

for the invention specifically provides for enhancement of low light condition images, night vision images, blurred images and other similar images.
Brief Description of the Drawing:
Figure 1 illustrates the image enhancement system according to the embodiment of the
invention
Figure 2 illustrates the pear shaped like mapping curve
Figure 3(a) illustrates an input image according the embodiment of the invention
Figure 3(b) illustrates an intermediate output image
Figure 3(c) illustrates an enhanced output image
Detailed Description:
For certain image processing applications, like object localization, the quality of the image, in terms of the contrast, brightness and sharpness plays a crucial role. The system of present invention provides for enhancing an input image by adjusting and improving the sharpness, brightness and the contrast of the image. In order to improve the contrast and brightness of the image, the system uses a pear shaped like curve for pixel value mapping. As the technique of the invention provides a visually enhanced image out of poor quality input image, it significantly reduces the further processing required for various image processing application thus improving upon the accuracy and outcome of the application for which it is used. As such, the technique of the present invention can be used as an effective pre-processing tool for various other image processing techniques and applications.
As illustrated in Figure 1, a poor quality image is input into the Input Module (100). The input image maybe an existing image, may be obtained run-time from a camera or other some other source. The Input Module (100) may comprise of various means like scanner. data cables, etc. The input image is converted into a gray scale image at the Input Module (100). The input image is then processed by a Generalized Space Frequency Representation (GSFR) Unit (105) and a Mapping Curve Unit (110) according to the embodiment of the invention to provide a visually enhanced output image. The enhanced output image is then displayed on the Output Module (115), which may be any display device known in the art, like LCD, monitor, LED, etc. Processing of the input image by

the GSFR Unit (105). helps de-noise the image while the Mapping Curve Unit (110) enhances an image selectively highlighting the object required for detection and providing an overall enhancement of the image.
In the GSFR Unit (105), Wigner distribution function and an exponential kernel are applied to the input image for image enhancement. Wigner distribution is a generalized time-frequency representation. Wigner distribution function is utilized for image processing, by extending it to a two-dimensional space. Utilization of the Wigner distribution function provides for a good quality of contrast and reduction of noise in the image. A four dimensional Wigner distribution function is used. The function has two space-domain variables 'x' and 'y' and two frequency-domain variables 'u' and 'v\ The Wigner distribution function is given by:

Where image size is MxN, window size is M' x N' 'f is the gray-scale function θ = 4π [uk /M + vl /N]
While applying the Wigner distribution function to an input image, V and 'y' represent the spatial coordinates of a candidate pixel. To improve the sharpness of the image and selectively reduce noise, an exponential kernel is applied along with the Wigner
distribution function and is calculated as Where, Lambda is 0.5.
With the exponential kernel, the Wigner distribution function is modified to:


The exponential kernel ensures that the candidate pixel (x, y) has maximum influence on the Wigner distribution function calculations and the influence rapidly decays out as one moves farther from the candidate pixel (x, y). Any other similar kernels known in the art, like cone shaped kernel, may be utilized with the Wigner distribution function.
In order to enhance the image range, the Wigner distribution function may be scaled by gain factor, α which can vary according to the need for degree of visual enhancement and depending on the quality of input image. Thus, the Wigner distribution function is further modified to:

The output image of the GSFR Unit (105) is further processed by the Mapping Curve Unit (110). The Mapping Curve Unit (110) comprises of application of a pear shaped like curve to the output image. The mapping curve is given by:

Where devout
y = output pixel value
x = input pixel value
a = maximum intensity value (G1 in Figure 2)

For a pear shaped like curve, the output gain increases as the input gray level value initially increases from zero; reaches up to the peak where output pixel value is equal to twice the pixel value of interest (Gp); and decreases gradually to zero after the peak. ]n the method of the invention, the pear shaped like curve is applied till the point where the output pixel value is equal to the input pixel value, after the peak. Further a linear equation is applied till the gray scale value 225. Finally, the values above 225 are clipped to 225. Accordingly, as illustrated in Figure 2, the pear shaped like curve is applied up to point 'Gl', where the input pixel value is equal to the output pixel value after the peak, i.e. Y-axis value is equal to 2*Gp, wherein 'Gp' is the pixel value of interest. A linear equation is applied from G1 to G2 where 'G2 = 225', which is fixed. By using this curve

according to the embodiment of the invention, the pixel values of interest will be mapped to the maximum intensity in an output image so as to get an enhanced visibility for the object of interest.
The pixel value of interest is selected depending on the object of interest which needs to be enhanced and hence may be application specific. For example. Figure 3(a) shows an N1R image of a parked vehicle captured at night time. In this case, vehicle is the object of interest and accordingly the pixel value of interest (Gp) is selected. In Figure 3(a) this pixel value of interest is, Gp=100 and GI=190 and G2=225. As such, the mapping curve is adjusted according to Gp=100. Figure 3(b) shows an intermediate output image by applying Wigner distribution function and the exponential kernel only, resulting in improved contrast, sharpness, brightness, de-noising of the image. Figure 3(c) shows an enhanced output image after applying the pear shaped like curve along with Wigner distribution function and the exponential kernel. As seen, the output image of Figure 3(c) is visually more enhanced than that of the intermediate output image of Figure 3(b) in terms of pixel value of interest ( that of vehicle's) being enhanced to the maximum intensity resulting in improved object enhancement. The visually enhanced output image is then displayed by the output module (115). Thus, depending on the object of interest in an image, pixel value of interest (Gp) will be changed accordingly in order to enhance the image.
The system of the present invention requires only one camera as opposed to the existing methods for image enhancement where multimodal imaging is required. Thus, the present invention provides for an image enhancement system which is simple, efficient and economical. The present invention is described in scientific terms using the mathematical formulae as stated herein. A person skilled in the art may appreciate that the values of these parameters are relative to application and do not limit the application of the invention.

We Claim,
1. A method and system for image enhancement using Generalized Space Frequency Representation wherein, the image enhancement technique, for use as a preprocessing tool, serving as an input to various other applications comprising an image enhancement process which is characterized in that the image enhancement is achieved by combining application of the Generalized Space Frequency Representation along with application of a mapping curve.
2. A method and system for image enhancement using Generalized Space Frequency Representation as claimed in Claim 1, wherein the application of the Generalized Space Frequency Representation comprises of application of the Wigner distribution and a an exponential kernel to the input image.
3. A method and system for image enhancement using Generalized Space Frequency Representation as claimed in Claim 1, wherein, application of the mapping curve leads to mapping of the pixel value of interest to the maximum intensity resulting in improved object enhancement.
4. A method and system for image enhancement using Generalized Space Frequency Representation as claimed in Claim 1, wherein pixel value of interest, as mapped by the mapping curve, is selected depending on the object of interest which needs to be enhanced.
5. A method and system for image enhancement using Generalized Space Frequency Representation as claimed in Claim 1, wherein, the application mapping curve is a pear shaped like curve.
6. A method and system for image enhancement using Generalized Space Frequency Representation as claimed in Claim 1, wherein, application of Generalized Space Frequency Representation improves the contrast, sharpness, brightness, de-noises the image and provides for a visually enhanced image.

7. A method and system for image enhancement using Generalized Space Frequency Representation as claimed in Claim 1, wherein, application of exponential kernel selectively reduces the noise and improves sharpness of an image.