Vignette Correction in Images

Guassian filter with higher value of Standard Deviation is used to remove vignette effect in images as it do averaging of intensity value.

Vignette Correction in Image

Shubhankar Sarkar

IIT Kharagpur

Electrical Engg. Department

Kharagpur,India

Abstract—Due to planar nature of the sensor and spherical nature of the image surface ,the image captured tend to be bright in the center and darker at the periphery. This effect is called vignetting. There has been proposed a very few number of method for vignette correction in image and maximum of them is unable to correct vignette in image totally. This paper introduces a guassian kernel to reduce the effect of vignette in image as it do averaging of intensity value. And it has been shown in this paper that very well guassian filter remove vignette effect in images.

Keywords—Vignette Correction , Correction of uneven illumination

I. Introduction

Images are important sources of information. There are many reason that degrade the quality of images. Out of them the effect of vignette in image play an impotant role. Vignette is the fall of pixel intensity from the centre towards the edge of images.Depending on the type of vignetting we distinguish several type of vignetting.

The causes of vegnetting are listed below.

Natural Vignetting refers to the light fall-off related to the geometry of the image forming sytem. It is usally described by cos law, which specifies the drop in light intensity depending on the angle formed between a ray of light entering the lens and the optical axis of lens.

Mechanical Vignetting refers to the light fall-off due to the light path blockage by elements of camera lens system,typically by an additional filter or hoods mounted on a lens .

Optical vignetting refers to the light fall-off caused by the blockage of off-axis incident light inside the lens body. The amount of blocked light depends on the physical dimensions of a lens.

Pixel vigenetting refers to the light fall-off related to the angular sensitivity of the image sensor pixel. It is caused by the pysical dimensions of a single pixel, in particular , the length of the ‘tunnel’ before the light reaches the photodiode.

The effect of vignetting on the image is undesirable in image processing and analysis, particularly in areas such as :image denoising , image segmentation , microscopic image analysic , visual surveilliance , sky image analysis , motion analysis in video sequences. Therefore it is very important to remove the effect of vidnette in images.

In this paper we propose a guassian filter to remove the effect of noise. The standard deviation of guassian filter is kept high to give smoothing output.

II. vignette correction using Guassian kernel

The kernel for smoothing , defines the shape of the function that is used to take the average of the neighboring points.A Guassian kernel is a kernel with the shape of a Guassian(normal distribution) curve.

Our method to correct vignette of a image is based on assumption that the uneven illumination is an additive low frequency signal. Therefore low pass filtering can be used to extract it from an image. The filtering may be achieved by convolving the image with a Guassian Kernel.The Guassian function G(x,y) is defined by

Where sigma define the effective spread of the Guassian function. The effect of this function is to delimit the spatial frequencies in an image , resulting in loss of edge definition and averaging of intensity values. The larger value of sigma which is nothing but the standard deviation of Guassian function will give more effect on smoothing. Approximating the shading in this manner is not a new idea but merely a recognized feature of Guassian function.

III. Results

Several input images are taken from Google. After applying these images through Guassian function we need to use image padding. There are different kind of image padding. Out of them ‘symmetric’ is taken, which is nothing but padding of image with mirror reflection of itself. Size of the guassian filter is specified as positive,odd integer or 2-element vector of positive,odd integers.

The results is given below when sigma or standard deviation of Guassian kernel is 15.

For low value of sigma (standard deviation of Guassian kernel) output is noisy, no smoothing is done and obviously vidnette effect is not removed.

The results is given below when sigma or standard deviation of Guassian kernel is 120.

Image-1

Image-2

When we increase the value of sigma upto120, the effect of vignette totally vanishes. And the smoothing effect of the filter increases.

Conclusion

In this paper the correction of vignette by Guassian filter is described. It is shown that Guassian kernel with lower value of sigma(Standard Deviation) is noisy and cannot remove the effect of vignette. But Guassian kernel with symmetric image padding and higher value of sigma become important because it can completely remove vignette of an image as it do averaging of intensity value.

References:

1. https://jcp.bmj.com/content/56/8/619
2. https://link.springer.com/article/10.1007/s11760-016-0941-2
3. https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.258.4780&rep=rep1&type=pdf