Priti Sehgal

Using PSO in a spatial domain based image hiding scheme with distortion tolerance

With tremendous advancement in digital technology, efficient steganography techniques are needed for the security and copyright protection of digital information being transmitted over the internet and for secret data communication. However, during transmission, a stego object may be exposed to noise or compression due to which the secret data cannot be extracted correctly at the receivers end. This paper presents an efficient spatial domain based image hiding scheme, using Particle Swarm Optimization (PSO). Here, PSO is used to find the best pixel locations in a gray scale cover image where the secret gray scale image pixel data can be embedded. The objective function for PSO is defined in such a way that both quality and robustness of the stego image are acceptable. This results in a stego image which is not only good in quality but is also able to sustain certain noise and compression attacks during transmission. The results when compared with some recent data hiding techniques in the spatial domain show better stego image quality along with distortion tolerance.

Fuzzy classification of pre-harvest tomatoes for ripeness estimation - An approach based on automatic rule learning using decision tree

Red-green color difference has the good classification capability than that from individual components R, G, B or red-green ratio in RGB color space.Adaptive segmentation technique segments the unripe tomato images from the background with a good effect.Decision trees for automatic rule learning repudiate the need of a human expert.Automatic fuzzy partitioning of the feature space to over linguistic terms.High true positive rate and lower false positive rate for the proposed fuzzy rule base classification. High classification accuracy achieved by the proposed system over learning algorithms. Tomato (Solanum lycopersicum) ripeness estimation is an important process that affects its quality evaluation and marketing. However, the slow speed, subjectivity, time consumption associated with manual assessment has been forcing the agriculture industry to apply automation through robots. The vision system of harvesting robot is responsible for two-tasks. The first task is the recognition of object (tomato) and second is the classification of recognized objects (tomatoes). In this paper, Fuzzy Rule-Based Classification approach (FRBCS) has been proposed to estimate the ripeness of tomatoes based on color. The two color depictions: red-green color difference and red-green color ratio are derived from extracted RGB color information. These are then compared as a criterion for classification. Fuzzy partitioning of the feature space into linguistic variables is done by means of a learning algorithm. A rule set is automatically generated from the derived feature set using Decision Trees. Mamdani fuzzy inference system is adopted for building the fuzzy rule based classification system that classifies the tomatoes into six maturity stages. Dataset used for experiments has been created using the real images that were collected from a farm. 70% of the total images were used for training and 30% images of the total were used for testing the dataset respectively. Training dataset is divided into six classes representing the six different stages of tomato ripeness. Experimental results showed the system achieved the ripeness classification accuracy of 94.29% using proposed FRBCS.

Using PSO in Image Hiding Scheme Based on LSB Substitution

With the massive growth in internet applications, there is a continuous need of efficient steganography techniques for the purpose of secret data communication and for the authentication and ownership identification of host data. This paper presents an efficient image hiding scheme using Particle Swarm Optimization (PSO) in the spatial domain of digital images. The proposed technique uses PSO to find the best pixel locations in an image where the secret image pixel data can be embedded. This PSO algorithm uses the Structural similarity Index (SSIM) as the objective function which is based on the simple visual effect of the human visual perception capability. As a result, the pixel positions generated by the proposed method, when used for embedding secret image data, result in minimum distortion of the host image. The results of the proposed technique have been analyzed qualitatively and quantitatively and also compared with some recent LSB techniques. The results show better stego image quality along with high embedding capacity.

Fingerprint Image Enhancement Using Type-2 Fuzzy Sets

High-quality images are a prerequisite for accurate matching of fingerprint images. However, fingerprint images are rarely of perfect quality. They may be degraded or corrupted due to variations in skin and impression conditions. Thus, fingerprint images must be enhanced before use. In this work, we demonstrate the efficacy of applying Type-2 fuzzy logic to fingerprint image enhancement when the input is pre-processed by the Hong’s algorithm. We have measured the fuzzy quality visually and quantitatively to note the effectiveness of the proposed technique for enhancement.

Image Retrieval Using Fuzzy Color Histogram and Fuzzy String Matching: A Correlation-Based Scheme to Reduce the Semantic Gap

The research interest in the recent years has progressed to improve the performance of image retrieval (IR) systems by reducing the semantic gap between the low-level features and the high-level concept. In this paper, we proposed an approach to combine the two modalities in IR systems, i.e., content and text, while considering the semantics between the query image and the textual query provided by the user. For content matching, color feature is extracted and is represented using fuzzy color histogram (FCH). For text matching, fuzzy string matching with edit distance is used. Furthermore, we find the correlation between the query image and the textual query provided by the user to reduce the semantic gap. Using this correlation, we combined the two modalities with late fusion approach. The proposed approach is assessed on standard annotated database. Higher values of precision and recall show better performance of the proposed approach. Moreover, the use of correlation helps in reducing the semantic gap and providing good results through better ranking of the similar images.

Securing Fingerprint Images Using PSO-Based Wavelet Domain Watermarking

ABSTRACT Watermarking techniques are used in biometric systems for the purpose of protecting and authenticating biometric data. This paper presents an efficient scheme to protect and authenticate fingerprint images by watermarking with their corresponding facial images in the wavelet domain using Particle Swarm Optimization (PSO). The key idea is to use PSO to find the best discrete wavelet transform (DWT) coefficients where the facial image data can be embedded. The objective function for PSO is based on the fingerprint image quality with respect to the Structural Similarity index (SSIM) and Orientation Certainty Level index (OCL). As a result, embedding the facial image data in the selected coefficients generated by the proposed method not only results in minimum distortion of the host image but also retains the feature set of the original fingerprint to be used in fingerprint recognition. The robustness of the watermark extracted using the proposed technique has been tested against various image processing attacks. This concept of watermarking a biometric image with another biometric image finds application in multimodal biometric authentication for a more secure system of personal recognition at the receivers end.

A Novel Framework for Enhancing Images Corrupted by Impulse Noise Using Type-II Fuzzy Sets

This paper presents a novel type-II fuzzy filter to remove impulse noise in an image. The filter processes impulses as type-II fuzzy sets. Type-II fuzzy sets model uncertainties more effectively than type-I fuzzy sets because the membership function for a type-I fuzzy set for a particular input is a crisp value. The proposed algorithm firstly detects impulses by considering grayscale distribution amongst neighbouring pixels and then determines the presence of impulsive pixels by comparing it with a range of threshold values using an S - shaped fuzzy membership function that is itself fuzzy. As the level of contamination varies from pixel to pixel, the modified value for the noisy pixel is calculated depending on the impulse noise present in it. The better performance of the filter is demonstrated on the basis of PSNR values calculated from the original and restored images respectively.

A novel approach to cartoon style rendering of an image with an approximated crayon texture

We present an efficient approach to cartoon style rendering of an image with an approximated crayon texture. The algorithm generates a crayon texture map, with two crayon textures (illuminated color texture and shadowed color texture) combined into one. The crayon texture is generated using crayon stroke generation technique. The texture is then mapped onto the image in cartoon style using per-vertex mapping concept. This algorithm results in an improved transition boundary between the two texture colors, comparable to the normal cartoon shaded image without crayon texture. If the image generated by this algorithm is animated, then the transition boundary would move logically with the object as compared to our earlier algorithm (described in the text), where transition boundary changed drastically with slight movement of an object in the image. The present approach can be used to generate variations in cartoon shading easily and efficiently. We have also attempted to approximate the effect of distribution of wax in the resulting image.

Multi-agent system for intelligent watermarking of fingerprint images

This paper presents a multi- agent system architecture for intelligent watermarking for securing fingerprint images. The proposed watermarking method uses a fuzzy-PSO based hybrid approach to secure a persons fingerprint image by watermarking it with its corresponding face image. As fingerprint databases are large, processing huge image data in real time is difficult. The proposed work uses a multi-agent system as a distributed system for performing watermarking of fingerprint images, where various subtasks are performed in parallel in the distributed system. For watermarking input fingerprints, each image is divided into blocks and type-2 fuzzy logic is used to calculate the watermarking strength of each block based on its features. Particle Swarm Optimization (PSO) is used to find optimum DCT coefficients of the image block to be used for watermark embedding in such a way that the quality and minutia matching ability of the host fingerprint are preserved. The proposed system WoFMAS (Watermarking of Fingerprints using Multi-Agent System) has multiple agents working on different blocks of the input fingerprint concurrently for efficiently distributing data and gathering results. The experimental study is done on the FVC 2004 fingerprint database and the results show that our hybrid approach gives better results in terms of watermarked image quality and robustness than other fuzzy based and other PSO based approaches in the literature.

Parallel weighted semantic fusion for cross-media retrieval

Efficiency and effectiveness of image retrieval IR system are based on the good interpretation and integration of multimodal information. Research results in the recent years show that combining the two modalities text based and content based even with simple fusion strategies alleviates the image retrieval results and also reduces the semantic gap. In this paper, we deploy parallel computing in weighted semantic similarity technique for IR using both text and content. This technique gives the weightage to the annotations associated with the query image based upon their semantic similarity with users query and then establishes the semantics with database images. The semantic similarity has been measured using WordNet. For content matching, colour feature is extracted and is represented using fuzzy colour histogram FCH. Furthermore, to fuse the two modalities, image reordering with late fusion strategy is used. Parallel processing is done at data level using the single program multiple data SPMD programming model that focuses on parallel execution of semantic similarity matching computations. The proposed approach shows that the parallel computing largely reduces the response time of the system. Whereas, semantics learned at an early stage helps in reducing the semantic gap. Experiments performed on two standard datasets reveal the good efficiency and effectiveness of the proposed approach.