Contrast Enhancement for Low Contrast Digital Images

Project Title

Contrast Enhancement for Low Contrast Digital Images

Project Area of Specialization Computer ScienceProject Summary

Digital images are usually of inferior quality and endure irregular lighting or illumination, failures of detail, and poor contrast. This becomes crucial when it is hard to differentiate the foreground of interest from the background, which makes the segmentation issue more severe and leads to misidentification. The fundamental thought of image enhancement is to build the contrast difference between light and dull areas to get better picture quality. The visual data of the picture will be increased to well explain and perceive, deliver vibrant images for the eyes, or help the features extraction process in the digital image processing. Digital image processing has a broad range of applications such as remote sensing, image, and data storage for transmission in business applications, medical imaging, acoustic imaging, Forensic sciences, and industrial automation.

Contrast Enhancement for Low Contrast Digital Image is a Desktop Application that works on the system and principles of Discrete wavelet transformation (DWT). The traditional contrast enhancement techniques enhance the image with a noisy and over-contrasted appearance. Previous Histogram equalization techniques did not work on the principles of DWT, However, we have proposed a hybrid method for the contrast enhancement to preserve the details of an image Moreover, the use of the fitness functions will enable us to reduce the time complexity and increase accuracy.   

Project Objectives

The objective of this project is the development of methodologies for image enhancement which is more accurate in terms of image quality assessment metrics and are less computationally intensive. The objectives for the proposal are given below:

1. To develop a system that will use to enhance low contrast images.
2. To develop a system that will act as a pre-processing module in any system that uses images as input and increases the image contrast before further processing the input images.
3. To develop a system for identifying any details in an image and magnifying that aspect for making it easier to extract useful information.
4. To develop a system that will assist any image domain expert in better understanding the output images.

Project Implementation Method

The proposed method uses single-level Discrete Wavelet Transformation (DWT). DWT divides the image into four components i.e. low-low (LL), low-high (LH), high-low (HL), and high-high (HH). The LL component is the approximate image to the input image used for the enhancement. The rest of the components have structural details which are kept constant so that noise is not enhanced. We apply CLAHE followed by Gamma Correction (GC) on the LL component. CLAHE enhances the overall contrast whereas GC controls the brightness and neutralizes the over-stretched artifacts caused by CLAHE. The fitness functions of clip limit and gamma value are designed for CLAHE and GC to optimize the enhancement. The enhanced LL component is integrated with LH, HL, and HH components. The reconstructed image is then brought back to the spatial domain by inverse DWT. The output is an enhanced image with better visual perception. It is observed that the proposed method enhanced the image without enhancing the noise which was a problem in the existing techniques. Moreover, we have improved results in terms of qualitative measures like Peak Signal-to-Noise Ratio (PSNR), Mean Squared Error (MSE), Structural Similarity Index (SSIM),  Root Average Squared Error (RASE), Spectral Angle Mapper (SAM), and Absolute Mean Brightness Error (AMBE).

Benefits of the Project

The software will provide an enhanced image of a low contrast digital image to the end-users. A lot of information is lost due to the poor quality of the camera and images captured under bad weather conditions which shows a lack of clarity in digital images. So, the software is intended to minimize all of those negative points and enhance the image to a better quality. The intended system is generic and can be used easily by any person with basic knowledge of the computer system. The software can help in the medical field, science and development sector which focuses on image data study, etc. The software will be used for medical imaging, e.g. in CT and MRI facilitating the diagnosis and detection of any tumor or disease. It will also support software development fields that deal specifically with image and data analysis through images. The system will also be helpful for the researchers and software engineers, to spot the desired results in an image.

Technical Details of Final Deliverable

Project Overview

Our project “Contrast Enhancement of Low Contrast Images” is a Desktop-based application that aims to provide the enhancement of low contrast images. In our software application, the users will be asked to input an image. The enhanced image will appear without noise enhancement and it will be more appealing than the original image. The traditional histogram equalization techniques did not work on the principles of DWT, However, we have proposed a hybrid method for contrast enhancement to preserve the details of an image.

Project features

1. Conversion of RGB images: In this feature, when a user inputs any type of image, the program checks if the image is in RGB or not, if it is then it proceeds to the next step, if not then it converts it into Gray level first before proceeding to the next step.
2. DWT-Processing: This feature converts the input image spatial domain to frequency domain, further dividing it into 4 components. After the conversion, it will apply contrast enhancement techniques on its main component which is a low-frequency component to enhance the image.
3. Contrast enhancement: In this step different contrast enhancement methods will be applied to the image as per user interest.
4.  Controlling Noise: This feature is related to the DWT-processing, as it is discussed earlier the frequency domain consists of 4 components, the noise will be controlled if the high-frequency components are separated i.e. HH, LH, and HL. These components contain different image noises detail.
5. Controlling Over-Enhancement: This feature is used to optimize the contrast enhancement to make the image look appealing. Sometimes an image looks so unnatural so that is why controlling over-enhancement will be used.
6. Image Detail information Preservation: This feature is used to make image detail information secured for instance when the image is enhanced along with noise then the enhanced image will not contain image detail as per the original image.
7. Reverse DWT: In this step the image which was divided into High-Frequency components and Low-Frequency components, then merged together for the conversion of the image into a spatial domain.

System Architect

The Architecture of the CELCI System has been divided into modules that are interlinked with each other. An image is decomposed into low-level frequency components and high-level frequency components. Low-level frequency components help our system to control over enhanced artifacts and on the other hand, high-level frequency components are kept constant to reduce noisy details of the original image. Then high-level frequency component will be merged with the low-level frequency component to form an image, here the image is in the frequency domain, and it will be converted back to the spatial domain image by applying reverse dwt. The system is designed in a way that, image details should be preserved.

Final Deliverable of the Project Software SystemCore Industry ITOther Industries Others Core Technology Artificial Intelligence(AI)Other Technologies OthersSustainable Development Goals Industry, Innovation and InfrastructureRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	20808
Google Collab Pro	Equipment	8	2601	20808