Understanding Image Intensities

By Berthold K. P. Horn. 1977

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Towards causal generative scene models via competition of experts (2020, background): “Proposed in the early days of computer vision Grenander (1976); Horn (1977), analysis-by-synthesis is an approach to the problem of visual scene understanding.”
Automatic differentiation in machine learning: a survey (2015, background): “problems is inverse graphics (Horn, 1977; Hinton and Ghahramani, 1997)—or analysis-bysynthesis (Yildirim et al.”
Medical Image Computing and Computer-Assisted Intervention - MICCAI 2003 (2003, methodology): “actuated stereoscopic endoscope was calibrated [13] and the accuracy of the displacements inside the surgical volume was evaluated to 4 pixels.”
Learning generative models of mid-level structure in natural images (2012, background): “While the processes leading to this percept are still poorly understood it is generally accepted that it cannot be formed based purely on the evidence available from the light falling on the retina, since this “inverse problem” is ill-posed (e.g. Poggio et al., 1985; see also Horn, 1977).”
Generative models for discovering sparse distributed representations. (1997, background): “In this paper, we take seriously the idea that vision is inverse graphics (Horn, 1977) and so we start with a stochastic, generative neural network that uses top-down connections to convert an abstract representation of a scene into an intensity image.”
The informed sampler: A discriminative approach to Bayesian inference in generative computer vision models (2014, background): “This view was advocated since the late 1970’ies (Horn, 1977; Grenander, 1976; Zhu and Mumford, 1997; Mumford and Desolneux, 2010; Mansinghka et al., 2013; Yuille and Kersten, 2006).”
Structure and motion from silhouettes (2001, methodology): “Examples of such techniques include shape from shading [58, 61, 142, 59, 145] under the assumptions of point light source and Lambertian surface, shape from line drawings [112, 50, 60, 28, 136, 83, 131, 122, 84, 102] under the assumption of trihedral-vertex polyhedral scene structure or smooth61”
Learning Stochastic Inverses (2013, background): “Posterior inference is the task of inverting a probabilistic model: Bayes’ theorem turns p(d|h) into p(h|d); vision is commonly understood as inverse graphics (Horn, 1977) and, more recently, as inverse physics (Sanborn et al.”
Three-Dimensional Computer Vision (1987, background): “Use of gradient image and reflectance map (Horn, 1977): (a) line drawing; (b) gradient image; (c) contours corresponding to the reflectance of three planes and a triangle with shape similar to (b)”
RECOVERING INTRINSIC SCENE CHARACTERISTICS FROM IMAGES (1978, background): “where I is intensity of incident illumination, R is reflectivity of the surface, and i is the angle of incidence of the illumination [ 20 ].”
The status of the minimum principle in the theoretical analysis of visual perception. (1985, methodology): “…developed accounts of edge detection (Marr & Hildreth, 1980), stereopsis (Grimson, 1981; Mart, 1982; Marr & Poggio, 1976, 1979) and of the perception of motion (Ullman, 1979), shape (Grimson, 1981; Horn, 1977; Ikeuchi & Horn, 1981), surface texture, brightness, lightness, and color (Marr, 1982).”
OpenDR: An Approximate Differentiable Renderer (2014, background): “It appears in the work of Grenander on analysis by synthesis [9], in physics-based approaches [13], in regularization theory [5,32], and even as a model for human perception [18,24,27].”