Jisha John

Enhancement of weather degraded video sequences using wavelet fusion

Under bad weather conditions, the contrast and colors of videos are degraded and it is imperative to include mechanisms that overcome weather effects from video sequences in order to make vision systems more reliable. Unfortunately it turns out that effects of weather cannot be overcome by simple image processing techniques. There are only few works and some existing methods in literature for enhancement of weather degraded video sequences. A novel approach is proposed in this paper that is used to enhance degraded video sequences. It enhances visibility of the frames and also maintains the color fidelity. First a background image is estimated for the video sequence. The enhancement method is then separately applied on this background image and on the estimated motion pixels. The enhancement method consists of three phases. The first phase estimates a global correction parameter and the second phase computes an approximate degradation measure. In the final phase a novel wavelet fusion method is used to obtain the enhanced frame. Performance analysis is carried out with the help of a contrast improvement index and sharpness measure. The method has been tested using real time video sequences and is found to give good results.

A review on cell detection and segmentation in microscopic images

Image analysis of cells and tissues is an important area of research, since it forms a basis for large number of biomedical applications. Many complexities such as heterogeneous shape of cells/nuclei, overlapping cells, presence of background noise, and variability in staining and illumination conditions make automated analysis of microscopic images, a complicated problem. Cell detection is the most basic and essential step for analysis of microscopic cell images. Even though many methods have been proposed over the years, ongoing research has failed to develop a system which is both generic and trainable for a wider range of applications with sensitivity and accuracy. Hence, novel automated methods are necessary. This paper examines and analyses various automated methods for cell detection and segmentation.

Detection and segmentation of mitotic cell nuclei in breast histopathology images

Detection and segmentation of cells is an important step for classifying the cells as cancerous or non-cancerous. Pathologists use microscopic images for analysis and further diagnosis of cancer. These images contain the microscopic structure of tissues and are stained using some staining components to facilitate the process. Staining process varies due to different stain manufacturers, staining practices and storage times. Manual methods are still used by pathologists in the detection of cancer. Automated methods assist the pathologists in early detection and diagnosis of cancer, and increase diagnostic precision. Even though many methods have been proposed over the years, ongoing research has failed to develop a system which is both generic and trainable for a wider range of applications with sensitivity and accuracy. Hence, novel automated methods are necessary. A robust, accurate and novel method is proposed for the detection and segmentation of cell nuclei, after which the cells are further classified as mitotic or non-mitotic.

Image processing techniques for surface characterization of nanostructures

The characterization of nanoscale images finds numerous applications in computer vision and image processing technologies. It is an emerging area of research and only few papers exist in literature on techniques for analyzing properties of nanostructures. The characterization of nanostructures in terms of surface morphology, particle size, porosity measurement etc helps in analyzing unique features and properties of the nanomaterials which makes them useful in many applications. This paper reviews the various algorithms and their advantages and drawbacks in analyzing the nanostructures.

Cell image segmentation and color coding based on nucleus to cell ratio

Biomedical analysis is a highly challenging area where lot of research activities is going on. Many automated biomedical image processing procedures have cell segmentation as its first step. Manual methods for this purpose are imprecise, tedious and highly subjective. Hence, novel automated methods are necessary. Analysis of cells includes segmenting the cells as well as computing the area of cell and nucleus. The nucleus to cellular ratio is crucial in determining the type of cells as well as detecting cancerous or damaged cells. In this paper, we propose a novel method to segment nucleus and cell and color code the cells based on their increasing nucleus to cellular ratio. Our tool will help scientists analyze biomedical images in a much better and efficient manner. The results of the proposed method have been compared with the most popular interactive image analysis tool ImageJ. The results obtained using our method is visually more appealing and easier for analysis.

Expectation-Maximization with Distance Measure for Color Image Segmentation

In this paper we propose an expectation-maximization (EM) algorithm with distance measure for color image segmentation. The probability distribution model used is the Gaussian mixture model. The concept of color distance measure is used in this algorithm to determine the region to which a particular pixel belongs. L *a* b color space is used to replace the more straightforward spaces such as the RGB color space and YUV color space. This algorithm is capable of automatically selecting the number of components of the model using minimum description length (MDL) criterion. The proposed method yields good segmentation with better PSNR and SSIM values compared to classical EM algorithm; that is, the segmented image will be structurally more similar to the original image.