MUMBAI, India, Feb. 13 -- Intellectual Property India has published a patent application (202641009165 A) filed by Dr. S. Ganeshmoorthy; J. Padma Priya; T. Nathiya; Ms. Aruna Devi K; Ms. Suganya Devaraj; Ms. J. Jobi Jesmitha; Dr M Jaithoon Bibi; Dr S Prasath; Dayananda Sagar Academy Of Technology & Manangement; and Dr. M. Amsaveni, Coimbatore, Tamil Nadu, on Jan. 29, for 'performance evaluation of particle swarm optimization based enhanced feature extraction algorithms in image tampering detection.'
Inventor(s) include Dr. S. Ganeshmoorthy; J. Padma Priya; T. Nathiya; Ms. Aruna Devi K; Ms. Suganya Devaraj; Ms. J. Jobi Jesmitha; Dr M Jaithoon Bibi; Dr S Prasath; and Dr. M. Amsaveni.
The application for the patent was published on Feb. 13, under issue no. 07/2026.
According to the abstract released by the Intellectual Property India: "Performance Evaluation of Particle Swarm Optimization based Enhanced Feature Extraction Algorithms in Image Tampering Detection Abstract: In today's world, image processing is always a concerned research area of interest. This is because of the significant role it played in the various fields like security, education etc. Image tampering is the major issue in the digital world. The tampered images appear with more sophistication. Increase in the availability of tampering tools, manipulation of digital multimedia contents is considered to be an easiest task. In this paper, feature extraction algorithms namely Local Binary Pattern and Speeded Up Robust Features are used to extract the required features from the images. Here, Enhanced Local Binary Pattern and Enhanced Speeded Up Robust Features are proposed to overcome the disadvantages in the existing algorithms. In this evaluation study, three common image classifiers such as Support Vector Machine, Ensemble and Back Propagation Neural Networks are used to classify the feature vectors. The particle swarm Optimization is also combined to optimize and choose the best one to obtain the better results. Experiments show that our proposed method ELBP has more efficient performance in accuracy and consumes very less time. The usage of digital images has been increasing exponentially with the inventions of latest models of smart phones and cameras. Social media is also adding is contribution to the image distribution. But the problem behind the collection of image is its reliability. Because the digital images can be tampered with the help of image editing tools and software very easily. The increasing effectiveness in image processing algorithms and the speed at which data is analyzed have made image processing less complex in the modern day. The detection of image manipulation is still a major problem in the computer vision sciences. There are numerous ways to get carried away with categorization and detection, but feature extraction plays the best. HOG is selected and used based on the quick robustness and rotation dependent representations. LBP extracts the feature vectors in histograms for each and every image. in order to these feature vectors to be used for direct classifier training . SURF characteristics offer robustness and good speed. It recognizes the transformation, size, and rotation of the item. It has its own stability to group points and creates object models. When paired with PSO, the algorithms to extract feature perform better and produce the best results. These algorithms serve as the most effective instrument for identifying and obtaining the important features. The features were optimized and by employing PSO with the specified feature extraction algorithms and results in good classification accuracy. The PSO algorithm can be used to solve issues involving both maximum and minimization. Even in the face of pre-processing and post-processing threats, PSO's performance is highly accepted. When it comes to improving the way input photos are grouped and assessing performance, classification is crucial. Every categorization algorithm has a unique purpose. It differs from every single appliance, for sure. Detection and classification can be done using Support Vector Machine. The number of features evaluates the performance of the classifier In ensemble classification, combining the decisions enhances the system's performance. Due to the inherent characteristics of the ensemble, many researches are conducted using ensemble Optimum initialization and efficient process is what BPNN is capable of. Comparing BPNN to other techniques yields better image quality and increased accuracy .The feature extraction algorithms extract the important features that provide the expected result and the classification methods provides the best match of image with database images."
Disclaimer: Curated by HT Syndication.
