Computer Vision

The improvement of vehicle security is a major priority of the car industry. Both active and passive security-systems have experienced a great development during the last decade, but the main research is still focused on minimising the errors committed by the driver rather than trying to avoid them. In this paper, a basic monocular vision system, focused on road location

ABSTRACT We present a computer vision system based on an integrated neural network architecture. In the low level vision subsystem, a network of networks-a biologically inspired network is used to recursively perform filtering, segmentation and edge detection; in the intermediate level and the high level, hierarchically structured arrays of self-organizing tree maps-extension of the popular self-organizing map are utilized to carry out image/feature analysis. The system has been applied to solve a number of real world problems. Some interesting and encouraging results are reported

ABSTRACT We present a computer vision system based on an integrated neural network architecture. In the low level vision subsystem, a network of networks-a biologically inspired network is used to recursively perform filtering, segmentation and edge detection; in the intermediate level and the high level, hierarchically structured arrays of self-organizing tree maps-extension of the popular self-organizing map are utilized to carry out image/feature analysis. The system has been applied to solve a number of real world problems. Some interesting and encouraging results are reported

This work contributes to vision processing for intelligent vehicle applications with an emphasis on Advanced Driver Assistance Systems (ADAS). A key issue for ADAS is the robust and efficient detection of free drivable space in front of the vehicle. To this end, we propose a stixel-based probabilistic color-segmentation algorithm to distinguish the ground surface from obstacles in traffic scenes. Our system learns color appearance models for free-space and obstacle classes in an online and self-supervised fashion. To this end, it applies a disparity-based segmentation, which can run in the background of the critical system path and at a lower frame rate than the color-based algorithm. This strategy enables an algorithm without a real-time disparity estimate. As a consequence, the current road scene can be analyzed without the extra latency of disparity estimation. This feature translates into a reduced response time from data acquisition to data analysis, which is a critical propert...

Sketch as an image search query is an ideal alternative to text in capturing the fine-grained visual details. Prior successes on fine-grained sketch-based image retrieval (FG-SBIR) have demonstrated the importance of tackling the unique traits of sketches as opposed to photos, e.g., temporal vs. static, strokes vs. pixels, and abstract vs. pixel-perfect. In this paper, we study a further trait of sketches that has been overlooked to date, that is, they are hierarchical in terms of the levels of detail -- a person typically sketches up to various extents of detail to depict an object. This hierarchical structure is often visually distinct. In this paper, we design a novel network that is capable of cultivating sketch-specific hierarchies and exploiting them to match sketch with photo at corresponding hierarchical levels. In particular, features from a sketch and a photo are enriched using cross-modal co-attention, coupled with hierarchical node fusion at every level to form a better ...

Sketch-based image retrieval (SBIR) is a cross-modal matching problem which is typically solved by learning a joint embedding space where the semantic content shared between photo and sketch modalities are preserved. However, a fundamental challenge in SBIR has been largely ignored so far, that is, sketches are drawn by humans and considerable style variations exist amongst different users. An effective SBIR model needs to explicitly account for this style diversity, crucially, to generalise to unseen user styles. To this end, a novel style-agnostic SBIR model is proposed. Different from existing models, a cross-modal variational autoencoder (VAE) is employed to explicitly disentangle each sketch into a semantic content part shared with the corresponding photo, and a style part unique to the sketcher. Importantly, to make our model dynamically adaptable to any unseen user styles, we propose to metatrain our cross-modal VAE by adding two style-adaptive components: a set of feature tr...

ABSTRACT In this paper, we present ErrorCheck, which is a new method for controlling the accuracy of a computer vision based pose refinement method. ErrorCheck consists of two parts: a method for validating the robustness of a pose refinement method towards false correspondences and a method for controlling the accuracy of a validated pose refinement method. ErrorCheck uses a theoretical estimate of the pose error covariance both for validating robustness and controlling the accuracy. We illustrate the first usage of ErrorCheck by applying it to state-of-the-art methods for pose refinement and some variations of these methods that turn out to be necessary to introduce.

Real-time vision is central to many embedded applications (e.g. vehicle guidance). It is a computationally intensive task well beyond current general purpose computing platforms such as PCs and workstations. Thus, most real time vision systems need special high performance computing platforms, commonly provided in the form of parallel processing engines or dedicated hardware. The proposed architecture uses new generations of re-programmable logic devices and modularised hardware, thereby gaining the performance advantage of hard-wired logic with the flexibility and associated economies of programmable systems. The architecture takes the form of an extensive processing hierarchy consisting of a set of tightly coupled parallel processors, each processing a portion of the image using a classic pipeline arrangement. A programmable image splitting (and reconstruction) engine feeds this array and offers the potential of further enhancing the performance of the engines by restructuring the...

Unmanned Air Vehicles (UAVs) have become an intelligent asset for surveillance, target tracking and reconnaissance in both urban and battlefield settings. This paper gives a framework for scale weight selection during feature extraction in aerial images from UAV. Dual-Tree Complex Waveform technique is used to extract rich feature descriptors of keypoints in images so that full phase and amplitude information can be retained at each scale. The scale weights are dependent on image characteristics such as the illumination and the contrast levels. The outcome of the framework shows promising results in terms of less redundancy of salient features from the images and hence improving the computational speed.

With the increasing number of cameras available in the cities, video traffic analysis can provide useful insights for the transportation segment. One of such analysis is the Automatic License Plate Recognition (ALPR). Previous approaches divided this task into several cascaded subtasks, i.e., vehicle location, license plate detection, character segmentation and optical character recognition. However, since each task has its own accuracy, the error propagation between each subtask is detrimental to the final accuracy. Therefore, focusing on the reduction of error propagation, we propose a technique that is able to perform ALPR using only two deep networks, the first performs license plate detection (LPD) and the second performs license plate recognition (LPR). The latter does not execute explicit character segmentation, which reduces significantly the error propagation. As these deep networks need a large number of samples to converge, we develop new data augmentation techniques that...

License plate recognition is an important task applied to a myriad of important scenarios. Even though there are several methods for performing license plate recognition, our approach is designed to work not only on high resolution license plates but also when the license plate characters are not recognizable by humans. Early approaches divided the task into several subtasks that are executed in sequence. However, since each task has its own accuracy, the errors of each are propagated to the next step. This is critical in the last two steps of the pipeline known as segmentation and recognition of the characters. Thus, we employ a technique to perform these two steps at once. The approach is based on a multi-task network where each task represents the recognition of an entire license plate character. We do not address the license plate detection problem in this paper. We also propose the use of a so called generative model for data augmentation of low-resolution images simulating ima...

In this study, a low cost, flexible and modular structure is proposed for mobile robot motion planning systems in an indoor environment with obstacles. In this system, the mobile robot has to follow the shortest path to the target avoiding obstacles. It is designed as three main modules including image processing, path planning and robot motion blocks. These modules are embedded on a single board computer. In the image processing module, the image of the indoor environment, including a mobile robot, obstacles and a target, having different colors is taken to the single board computer with a wireless IP camera. This image is processed to find the locations of the mobile robot, obstacles and the target in C programming language using OpenCV. In path planning module, the shortest and optimal path is generated for the mobile robot. Generated path is applied to the robot motion module to produce necessary angles and distances for the mobile robot to reach the target. Since the structure ...