Original research articleRestoration and quality improvement of distorted tribal artworks using Particle Swarm Optimization (PSO) technique along with nonlinear filtering
Introduction
Art has the power to influence people and bring people closer. Art may vary from culture to culture yet it holds the key to bring communities closer. These artworks are storehouse of information about various cultures across the globe. Though in modern times, such cultural significant artworks fail to get much attention or recognition and are usually not properly preserved. These can be preserved through digitisation and those artworks which have lost its radiance over time can be reconstructed or restored using modern technology. In this paper, one such tribal artworks of India have been used as the sample images. The aim of this work is to restore and improve the quality of these artworks that are degraded over time, PSO (Particle Swarm Optimization) technique have been used in this paper. PSO was introduced by Kennedy and Eberhart [1] which was inspired by the motion of a flock of birds. Various such optimization techniques have been proposed over the years, some of which include Ant Colony Optimization (ACO) [2], [3], [4], Differential Evolution (DE) [5], Bacterial Foraging Optimization (BFO) [6], [7], [8], [9], Artificial Bee Colony (ABC) [10], [11], [12], Glow-worm Swarm Optimization (GSO) [13], [14], Bat Algorithm (BA) [15], [16] and more. The two most famous from the above mentioned ones are ACO and PSO. These optimization algorithms are used in different problems related optimization in numerous fields [17], [18], [19]. Considering the image enhancement as an optimization problem allows us to use these techniques for image enhancement as well [20], [21], [22], [23]. To improve the quality of image, a parameterised transformation function [23] is used in this paper, in which the values of the parameters are optimised by PSO technique with the help of an objective function. In this paper, a combination of median Filter and PSO is used for improving the image quality of old and faded colour images. MATLAB software has been used to develop the program. The results obtained are compared with that of other techniques like CLAHE, BBHE and DSIHE [24], [25], [26].
Section snippets
Median filter
It is an example of nonlinear filter i.e., output of median filter is a nonlinear function of its input. Linear filter can’t provide edge preservation, while the nonlinear filter can. The median filter replaces the pixel value by calculating the median of the neighbouring pixels. The neighbouring pixels window size is selected by the user. The activity of a median filter can be represented by the following mathematical equation [27],Here w represents the size of the
Absolute Mean Brightness Error (AMBE)
It represents the change in absolute mean value of the original and filtered image [36]. This value indicates the brightness preservation rate during the filtering process. Smaller AMBE value is desired because it signifies better brightness preservation.where, represents original image and is the restored image.
Peak to Signal Noise Ratio (PSNR)
It is a ratio between the maximum signal power to the distorting noise power and is represented in dB [37]. High PSNR value indicates better image filtering. If
Results and discussions
The necessary program for median filtering followed by PSO is developed in MATLAB and is tested on various distorted colour images, results of one such image is presented in this paper. The result obtained from the proposed method is compared with different histogram equalization techniques namely, CLAHE, BBHE and DSIHE. All the mentioned methods’ performance is compared on the basis of image quality parameters like PSNR, AMBE, NMSE, MSE and CPP. Fig. 2(a) shows the distorted and degraded input
Conclusion
A PSO based image restoration method is presented in this paper which pre-processes the image using median filter to remove noise and to give better results. Results are compared with other techniques like CLAHE, BBHE and DSIHE. On comparison, it is found that the method involving PSO gives the best results. Noises of the distorted images are removed using nonlinear median filter. It works well in preserving edges of the input images as well. The corresponding program is developed using MATLAB
Declaration of Competing Interest
The authors certify that there is no conflict of interest with any other.
Acknowledgement
The authors would like to acknowledge Prof. R.C. Jha, Dr. R.K. Sarkar, Dr. S. Karmakar and Birla Institute of Technology, Mesra for the support provided to carry out this work. This work is financially supported by Indian National Science Academy (INSA).
References (38)
- et al.
Gaussian mixture model for texture characterization with application to brain DTI images
J. Adv. Res.
(2019) - et al.
Multifractal analysis of ceramic pottery SEM images in Cucuteni-Tripolye culture
Optik
(2018) - et al.
Gaussian mixture model for texture characterization with application to brain DTI images
J. Adv. Res.
(2019) - et al.
Optical pressure sensors based plantar image segmenting using an improved fully convolutional network
Optik
(2019) - et al.
Investigation on quality enhancement of old and fragile artworks using non-linear filter and histogram equalization techniques
Optik
(2021) - J.Kennedy, R.Eberhart, Particle Swarm Optimization, in: IEEE Proceedings of ICNN'95 International Conference on Neural...
Optimization, Learning and Natural Algorithms (Ph.D. thesis)
(1992)- et al.
Blood pressure and flow values in small vessels angioarchitectures: application for diabetic retinopathy
Rom. J. Phys.
(2016) - et al.
Towards accurate diagnosis of skin lesions using feed forward back propagation neural networks
Diagnostics
(2021) - et al.
Differential evolution–a simple and efficient heuristic for global optimization over continuous spaces
J. Glob. Optim.
(1997)
Digital Image Processing Using SCILAB
An efficient local binary pattern based plantar pressure optical sensor image classification using convolutional neural networks
Optik
Biomimicry of bacterial foraging for distributed optimization and control
IEEE Control Syst. Mag.
Intensity-based classification and related methods in brain MR images
Classif. Clust. Biomed. Signal Process.
Edge-based structural similarity analysis in brain MR images
J. Med. Imaging Health Inform.
A new metaheuristic bat-inspired algorithm
Nature Inspired Cooperative Strategies for Optimization
Cited by (10)
Application of Retinex and histogram equalisation techniques for the restoration of faded and distorted artworks: A comparative analysis
2023, OptikCitation Excerpt :Taking inspiration from this, image enhancement techniques named Single Scale Retinex (SSR), Multi Scale Retinex (MSR) and MSR with Colour Restoration (MSRCR) are developed [14–19]. Apart from these different other techniques for the restoration and enhancement of distorted image quality, have been reported by the researchers [20–33]. The performance of HE techniques like CLAHE, BBHE, DSIHE and retinex theory based methods like SSR, MSR, MSRCR are compared on the basis of image quality parameters, such as PSNR, AMBE, NMSE, CPP, SSIM, r and IEF [34,35].
A quasi-reflection based SC-PSO for ship path planning with grounding avoidance
2022, Ocean EngineeringCitation Excerpt :A linguistic information granulation penalty function was designed based on co-evolutionary particle swarm optimization algorithm (Zhang et al., 2021). A restoration and quality improvement method of distorted tribal artworks was designed based on particle swarm optimization (Kaur and Dutta, 2021). Particle swarm optimization algorithm has the advantage of using its previous experience and the experience of other social members to adjust its behavior.
Comparative performance analysis of Fuzzy Logic and Particle Swarm Optimization (PSO) techniques for image quality improvement: With special emphasis to old and distorted folk paintings
2022, OptikCitation Excerpt :The relation can be positive or negative [33]. The fuzzy technique and PSO algorithm [25] were implemented for colour image using MATLAB and the result obtained for the same are discussed in this paper. Both of the mentioned methods were compared with the help of parameters like PSNR, AMBE, NMSE, CPP, SSIM, r and IEF.
Improving Significant Wave Height Prediction Using a Neuro-Fuzzy Approach and Marine Predators Algorithm
2023, Journal of Marine Science and EngineeringMetaheuristic-based PID Control: Evaluating the Effect of Parameter Settings
2023, Research Square