B. Radhakrishnan

Tumor region extraction using edge detection method in brain MRI images

In todays modern medical arena patients with brain tumor are increasing rapidly with a fast pace and above the par. Detection of brain tumor has become a challenging task to compete with. In this paper an automated method for detecting brain abnormalities and tumor edema has been proposed using sobel edge detection method. Various MRI images have been used as inputs here. Here, first of all the pre-processing of image has been done to cut out any discrepancy in it and then the image has been smoothened using median filter. We have proposed an appropriate method to find threshold value using standard deviation and we get an intensity map. Now we recomputed standard deviation for this intensity map. Using this we will calculate an average intensity of the pixels those are above this standard deviation. Finally, this computed average intensity will be taken as the threshold value to segment the tumor from the original MRI images. The intensity value greater than and equal to the calculated threshold value is set to 255 and less than is set to 0, this segments our abnormal region which is tumor. At last, we use sobel edge detector to identify the border of the tumor region. The outcome of the proposed method improves efficacy and accuracy for detection of brain tumors.

Detection and extraction of roads from satellite images based on Laplacian of Gaussian operator

The extraction of road networks from satellite images has fundamental importance in GIS applications. In this paper, an automatic approach for road extraction is proposed to extract the road components from satellite images using Laplacian of Gaussian operator. The image is first pre-processed to identify the color space components and then smoothened using Laplacian of Gaussian. Then morphological method is used to remove the unwanted objects in the image. The result of proposed method improves efficiency and accuracy for extraction of roads.

Face recognition on surgically altered faces using principal component analysis

The popularity of plastic surgery is increasing due to the enhancement and accessibility in that field. Plastic surgery procedure provides an efficient way to improve the facial appearance, which result in the reconstruction of facial features either globally or locally. The changes introduced by plastic surgery remains complicated to be analyzed by face recognition systems. This paper focus on recognition of face images before and after plastic surgery. Here, Viola-Jones algorithm is used to detect the face and modified PCA algorithm is used to recognize faces from images with some near real-time variations. This principal component analysis resolves the recognition problem for 2-D image of faces, using Eigenface Recognition algorithm.

Automatic number plate localization using dynamic thresholding and morphological operations

The increase in accidents, vehicle thefts, over speeding and traffic infringements make proper traffic rule enforcement a necessity. Speed traps play an important role in reducing the rate of accidents at accident prone areas. Automation of number plate recognition system makes traffic rule enforcement an easy task without direct human intervention. In this paper we propose a less complex and effective system to detect and segment the number plate region. The method comprises of a dynamic thresholding technique, morphological operations followed by connected component analysis. Dynamic thresholding helps the system to detect the number plate region at varying illumination conditions. Morphological operations eliminate the unwanted small objects other than the number plate and connects the disconnected edges.

Developing rumor identification analysis on online social network with efficient cure search

Social sites have emerged as an ideal platform for content sharing and searching in modern era. With the exponential growth of users, social media has gained more attention than mass media. Social media serve as a discussion panel for authenticated users. This offers users to share their contents on social media. But, it may be either genuine or not. In such circumstances, there is a possibility of spreading false information among peoples. This may affect the trustworthiness of users about social media. Hence, it is essential to check the veracity of rumor in social media. So, we propose some strategies for rumor identification by analyzing the user behaviours. Beyond this, for enhancing the efficiency and accuracy of traditional searching in social media, we propose a time consuming searching strategy called CURE search. CURE search is dynamic attribute search in which exact match can be found out from similar profiles from catalogue with better speed.

A Graphical Password Authentication for analyzing legitimate user in online social network and secure social image repository with metadata

Internet plays a crucial role in todays life, so the usage of online social network monotonically increasing. People can share multimedia informations fastly and keep in touch or communicate with friends easily through online social network across the world. Security in authentication is a big challenge in online social network and authentication is a preliminary process for identifying legitimate user. Conventionally, we are using alphanumeric textbased password for authentication approach. But the main flaw points of text based password is highly vulnerable to attacks and difficulty of recalling password during authentication time due to the irregular use of passwords. To overcome the shortcoming of text passwords, we propose a Graphical Password authentication. An approach of Graphical Password is an authentication of amalgam of pictures. It is less vulnerable to attacks and human can easily recall pictures better than text. So the graphical password is a better alternative to text passwords. As the image uploads are increasing by users share through online site, privacy preserving has become a major problem. So we need a Caption Based Metadata Stratification of images for delivers an automatic suggestion of similar category already in database, it works by comparing the caption metadata of album with caption metadata already in database or extract the synonyms of caption metadata of new album for checking the similarity with caption metadata already in database. This stratification offers an enhanced automatic privacy prediction for uploaded images in online social network, privacy is an inevitable factor for uploaded images, and privacy violation is a major concern. So we propose an Automatic Policy Prediction for uploaded images that are classified by caption metadata. An automatic policy prediction is a hassle-free privacy setting proposed to the user.

Intelligent parking space detection and number plate extraction

Automatic parking space detection decreases the time wasting to find a proper parking space. The proposed work helps the parking management system to detect an empty parking lot and allocate that space to a particular vehicle. The vehicle number can be used to identify vehicle parked in which lot. Sobel edge detection method is used to find empty parking lot and for number plate extraction morphological operations such as top hat, dilation and erosion are used.

Detecting tail lights for analyzing traffic during night using image processing techniques

Recent studies show that death rate is increasing day by day because of road accidents. Analysis of traffic is a major problem in roads. Advanced Driver Assistance System (ADAS) is the technique implemented to avoid traffic accidents. These methods are all available for only day time conditions. Vehicle detection during night is a major challenge now days. During night the environmental conditions changes, reflection of street lights etc., deviates the detection of vehicle light. This may cause accidents on larger account. Many image processing techniques has been implemented to detect vehicle by analyzing the tail light during night.

Compression of map images using code book method

Map images are widely used in variety applications such as personal navigation, mobile computing, internet and academia. The usage of raster maps in many applications especially in small storage devices has negative impacts because of its relatively large size. The efficient use of compression method is essential to meet such requirements. A fast lossless compression scheme for raster map images is proposed in this paper. The proposed scheme is a combination of two methods, Code book construction based on indexing and Row-Column Reduction coding Algorithm. The code book based on indexing is constructed from the similar blocks present in the map image and row column reduction coding algorithm is used to compress the unique blocks in the image. Our experimental results show that the proposed compression scheme has better performance and the scheme is simple and fast.

Extraction of fire region from forest fire images using color rules and texture analysis

Day by day the rate of forest fire is increasing and the losses suffered are huge. There are many methods available for detecting the fire flames. However most of them fail in discriminating fire and fire colored local objects. In this paper we propose a novel and robust method that segments fire flame regions from an image. The proposed method combines the unique color and texture features of the fire flames. The RGB images obtained from digital cameras are used as inputs. The mean value estimation is used to assume the presence of fire flames in an image. Its presence is confirmed by performing the background elimination. Color rules defined for the RGB, HSV, and YCbCr color spaces are used to segment the fire regions. The HSV and YCbCr color spaces are used because of their capability to separate intensity (luminance) and color characteristics (chrominance). One of the unique characteristics of the fire flame pixels is its high intensity value. These color spaces helps to take this characteristic into account thereby reducing a lot of errors. Finally the result is made more accurate by using the texture analysis. The Gray-Level Co-occurrence Matrix (GLCM) is used to calculate five texture features of the fire flames. The proposed method is tested on a set of fire and non-fire images and it achieves a high detection rate and low false rate.