Abdul Basit Shaikh

Face recognition using weighted distance transform

Face recognition is one of the most important computer vision problems. Its importance is largely due to the security issues the world is facing at the moment and also the requirement of a more robust system security standard. This work investigates the use of facial weighted distance transform to improve the face recognition rate. Weighted distance transform, also known as geodesic distance, not only considers the spatial distance among pixels but also takes into account the local intensity variations providing a distance transform in the spatiointensity domain. Geodesic distance transform of facial images is estimated using the “Fast Marching” [1, 2] technique which is based on Dijkstras algorithm employed to identify the shortest network path. It is a single pass algorithm providing efficient geodesic distance feature vector, thereby reducing the recognition time. A standard Frontal Face Data Base [3] is used to validate the algorithm. The obtained results are comparable to the state-of-the-art face recognition techniques.

“Cellular-Cut”-Interactive n-Dimensional Image Segmentation Using Cellular Automata

Interactive segmentation of images has become an integral part of image processing applications. Several graph based segmentation techniques have been developed, which depend upon global minimization of the energy cost function. An adequate scheme of interactive segmentation still needs a skilled initialization of regions with user-defined seeds pixels distributed over the entire image. We propose an iterative segmentation technique based on Cellular Automaton which focuses to reduce the user efforts required to provide initialization. The existing algorithms based on Cellular Automaton only use local smoothness term in label propagation making them highly sensitive to user-defined seeds pixels. To reduce the sensitivity towards initial user definition of regions, global constraints are introduced along with local information to propagate labels. The results obtained are comparable to the state-of-the-art interactive segmentation techniques on a standard dataset.

A New Framework for Interactive Images Segmentation

Image segmentation has become a widely studied research problem in image processing. There exist different graph based solutions for interactive image segmentation but the domain of image segmentation still needs persistent improvements. The segmentation quality of existing techniques generally depends on the manual input provided in beginning, therefore, these algorithms may not produce quality segmentation with initial seed labels provided by a novice user. In this work we investigated the use of cellular automata in image segmentation and proposed a new algorithm that follows a cellular automaton in label propagation. It incorporates both the pixels’ local and global information in the segmentation process. We introduced the novel global constraints in automata evolution rules; hence proposed scheme of automata evolution is more effective than the automata based earlier evolution schemes. Global constraints are also effective in deceasing the sensitivity towards small changes made in manual input; therefore proposed approach is less dependent on label seed marks. It can produce the quality segmentation with modest user efforts. Segmentation results indicate that the proposed algorithm performs better than the earlier segmentation techniques.

A Global Sampling Based Image Matting Using Non-Negative Matrix Factorization

Image matting is a technique in which a foreground is separated from the background of a given image along with the pixel wise opacity. This foreground can then be seamlessly composited in a different background to obtain a novel scene. This paper presents a global non-parametric sampling algorithm over image patches and utilizes a dimension reduction technique known as NMF (Non-Negative Matrix Factorization). Although some existing non-parametric approaches use large nearby foreground and background regions to sample patches but these approaches fail to take the whole image to sample patches. It is because of the high memory and computational requirements. The use of NMF in the proposed algorithm allows the dimension reduction which reduces the computational cost and memory requirement. The use of NMF also allow the proposed approach to use the whole foreground and background region in the image and reduces the patch complexity and help in efficient patch sampling. The use of patches not only allows the incorporation of the pixel colour but also the local image structure. The use of local structures in the image is important to estimate a high-quality alpha matte especially in the images which have regions containing high texture. The proposed algorithm is evaluated on the standard data set and obtained results are comparable to the state-of-the-art matting techniques.

A global non-parametric sampling based image matting

Image matting is a process of separating the foreground objects from an image along with opacity values for each pixel. It is an under-constraint problem hence a user interaction is required to identify the definite foreground, background and semi-transparent pixels. In general the information in the definite foreground and background regions is modeled locally to estimate the foreground and background color of the pixels in the semi-transparent region which are then used to estimate opacity values. A global non-parametric sampling based approach is presented which incorporates not only the color information in the foreground and background regions but also utilizes the local structure of an image to improve the quality of the estimate matte. This global sampling approach reduces the segmentation mis-classification that is incorporated in the resulting alpha matte by considering only the local color information. The results obtained are comparable to the state of the art image matting techniques on a standard dataset.