Orientation (computer vision)

About: Orientation (computer vision) is a research topic. Over the lifetime, 17196 publications have been published within this topic receiving 358181 citations.

Information limit on the spatial integration of local orientation signals.

Abstract: Channel-based models of human spatial vision require that the output of spatial filters be pooled across space. This pooling yields global estimates of local feature attributes such as orientation that are useful in situations in which that attribute may be locally variable, as is the case for visual texture. The spatial characteristics of orientation summation are considered in the study. By assessing the effect of orientation variability on observers' ability to estimate the mean orientation of spatially unstructured textures, one can determine both the internal noise on each orientation sample and the number of samples being pooled. By a combination of fixing and covarying the size of textured regions and the number of elements constituting them, one can then assess the effects of the texture's size, density, and numerosity (the number of elements present) on the internal noise and the sampling density. Results indicate that internal noise shows a primary dependence on texture density but that, counterintuitively, subjects rely on a sample size approximately equal to a fixed power of the number of samples present, regardless of their spatial arrangement. Orientation pooling is entirely flexible with respect to the position of input features.

Omnidirectional bar code reader with method and apparatus for detecting and scanning a bar code symbol

Abstract: A bar code reader includes an optical system for storing a two dimensional image in memory, which stored two dimensional image contains a bar code symbol A method and apparatus is disclosed for detecting the location of said bar code image by computing the accumulated sum of the products of the derivatives of respective first and second scan lines as a location score for the image under consideration The higher the location score, the higher the probability that the area under consideration contains a bar code image Also, a method and apparatus is disclosed for determining the fine orientation of a located bar code image by the cross-correlation of interpolated scan line data The bar code image is filtered by shifting interpolated scan line data in accordance with the detected peak of the cross-correlation and combining the shifted scan line data

ReDet: A Rotation-equivariant Detector for Aerial Object Detection

Abstract: Recently, object detection in aerial images has gained much attention in computer vision. Different from objects in natural images, aerial objects are often distributed with arbitrary orientation. Therefore, the detector requires more parameters to encode the orientation information, which are often highly redundant and inefficient. Moreover, as ordinary CNNs do not explicitly model the orientation variation, large amounts of rotation augmented data is needed to train an accurate object detector. In this paper, we propose a Rotation-equivariant Detector (ReDet) to address these issues, which explicitly encodes rotation equivariance and rotation invariance. More precisely, we incorporate rotation-equivariant networks into the detector to extract rotation-equivariant features, which can accurately predict the orientation and lead to a huge reduction of model size. Based on the rotation-equivariant features, we also present Rotation-invariant RoI Align (RiRoI Align), which adaptively extracts rotation-invariant features from equivariant features according to the orientation of RoI. Extensive experiments on several challenging aerial image datasets DOTA-v1.0, DOTA-v1.5 and HRSC2016, show that our method can achieve state-of-the-art performance on the task of aerial object detection. Compared with previous best results, our ReDet gains 1.2, 3.5 and 2.6 mAP on DOTA-v1.0, DOTA-v1.5 and HRSC2016 respectively while reducing the number of parameters by 60% (313 Mb vs. 121 Mb). The code is available at: https://github.com/csuhan/ReDet.

A minimal case solution to the calibrated relative pose problem for the case of two known orientation angles

Abstract: It this paper we present a novel minimal case solution to the calibrated relative pose problemusing 3 point correspondences for the case of two known orientation angles. This case is relevant when a camera is coupled with an inertial measurement unit (IMU) and it recently gained importance with the omnipresence of Smartphones (iPhone, Nokia N900) that are equippedwith accelerometers tomeasure the gravity normal. Similar to the 5-point (6-point), 7-point, and 8-point algorithm for computing the essential matrix in the unconstrained case, we derive a 3-point, 4-point and, 5-point algorithm for the special case of two known orientation angles. We investigate degenerate conditions and show that the new 3-point algorithm can cope with planes and even collinear points. We will show a detailed analysis and comparison on synthetic data and present results on cell phone images. As an additional application we demonstrate the algorithm on relative pose estimation for a micro aerial vehicle's (MAV) camera-IMU system.

Partial shape recognition using dynamic programming

Abstract: A partial-shape-recognition technique utilizing local features described by Fourier descriptors is introduced. A dynamic programming formulation for shape matching is developed, and a method for comparison of match quality is discussed. This technique is shown to recognize unknown contours that may be occluded or that may overlap other objects. Precise scale information is not required, and the unknown objects may appear at any orientation with respect to the camera. The segment-matching dynamic programming method is contrasted with other sequence-comparison techniques that utilize dynamic programming. Experimental results are discussed that indicate that partial contours can be recognized with reasonable accuracy. >