Shanti K. Guru
Bio: Shanti K. Guru is an academic researcher from College of Engineering, Pune. The author has contributed to research in topic(s): Digital image processing & Image restoration. The author has an hindex of 1, co-authored 1 publication(s) receiving 3 citation(s).
••03 Sep 2015
TL;DR: An algorithm was proposed to synthesize the structure & texture as well as fill the hole that is left behind in an undetectable form of image inpainting, and an attempt has been made to compute actual color values using exemplar based texture synthesis and region filling method.
Abstract: Image inpainting is an interesting new research topic in image processing, which can be used in many thrust areas, such as computer graphics, image editing, film postproduction, image restoration and special effects and the restoration of old Photographs and damaged film, removal of superimposed text like dates, subtitles, or publicity and the removal of entire objects from the image. In image inpainting, missing (target) regions were filled by structural and textural information of an image in a visually plausible way, also known as image restoration. Though this technique is very useful, it is still a challenging problem in computer vision and computer graphics. In this paper, an algorithm is proposed for removing target objects from digital images. An algorithm was proposed to synthesize the structure & texture as well as fill the hole that is left behind in an undetectable form. An attempt has been made to compute actual color values using exemplar based texture synthesis and region filling method. A number of examples on removing occluding objects from real and satellite images demonstrate the effectiveness of proposed algorithm in terms of both inpainting quality and computational efficacy.
TL;DR: Results prove the efficiency of the proposed generalized scanline fill algorithm and its advantage over the state-of-the-art algorithms, and clearly show that optimized machine routine is capable of performing the real-time tasks.
Abstract: A generalized iterative scanline fill algorithm intended for use in real-time applications and its highly optimized machine code implementation are presented in this paper. The algorithm uses the linear image representation in order to achieve the fast memory access to the pixel intensity values. The usage of the linear image representation is crucial for achieving the highly optimized low-level machine code implementation. A few generalization features are also proposed, and discussion about the possible real-time applications is given. The proposed efficient machine code implementation is tested on several PC machines, and a set of numerical results is provided. The machine routine is compared with standard and optimized implementations of the 4-way flood fill algorithm and scanline fill algorithm. The machine code implementation performs approximately 2 times faster than the optimized scanline fill algorithm implementation and 6 times faster than standard iterative scanline fill algorithm implementation on two-dimensional image data structure. Furthermore, the machine routine proved to perform even more than 15 times faster than the optimized flood fill algorithm implementations. Provided results prove the efficiency of the proposed generalized scanline fill algorithm and its advantage over the state-of-the-art algorithms, and clearly show that optimized machine routine is capable of performing the real-time tasks.
•01 Jul 2020
TL;DR: The applicability of the setting and processing pipeline on affective state prediction based on front camera recordings during math-solving tasks and emotional stimuli from pictures shown on a tablet are demonstrated.
Abstract: Front camera data from tablets used in educational settings offer valuable clues to student behavior, attention, and affective state. Due to the camera’s angle of view, the face of the student is partially occluded and skewed. This hinders the ability of experts to adequately capture the learning process and student states. In this paper, we present a pipeline and techniques for image reconstruction of front camera recordings. Our setting consists of a cheap and unobtrusive mirror construction to improve the visibility of the face. We then process the image and use neural inpainting to reconstruct missing data in the recordings. We demonstrate the applicability of our setting and processing pipeline on affective state prediction based on front camera recordings (i.e., action units, eye gaze, eye blinks, and movement) during math-solving tasks (active) and emotional stimuli from pictures (passive) shown on a tablet. We show that our setup provides comparable performance for affective state prediction to recordings taken with an external and more obtrusive GoPro camera.
TL;DR: The experimental results show that the proposed methods are capable of reconstructing missing regions with good visual quality.
Abstract: The Consultative Committee for Space Data Systems proposed an efficient image compression standard that can do lossless compression (CCSDS-ICS). CCSDS-ICS is the most widely utilized standard for satellite communications. However, the original CCSDS-ICS is weak in terms of error resilience with even a single incorrect bit possibly causing numerous missing pixels. A restoration algorithm based on the neighborhood similar pixel interpolator is proposed to fill in missing pixels. The linear regression model is used to generate the reference image from other panchromatic or multispectral images. Furthermore, an adaptive search window is utilized to sieve out similar pixels from the pixels in the search region defined in the neighborhood similar pixel interpolator. The experimental results show that the proposed methods are capable of reconstructing missing regions with good visual quality.