This paper presents a literature survey on existing disparity map algorithms. It focuses on four main stages of processing as proposed by Scharstein and Szeliski in a taxonomy and evaluation of dense two-frame stereo correspondence algorithms performed in 2002. To assist future researchers in developing their own stereo matching algorithms, a summary of the existing algorithms developed for every stage of processing is also provided. The survey also notes the implementation of previous software-based and hardware-based algorithms. Generally, the main processing module for a software-based implementation uses only a central processing unit. By contrast, a hardware-based implementation requires one or more additional processors for its processing module, such as graphical processing unit or a field programmable gate array. This literature survey also presents a method of qualitative measurement that is widely used by researchers in the area of stereo vision disparity mappings.

/pdf/literature-survey-on-stereo-vision-disparity-map-algorithms-3mtr9e2zyc.pdf

Literature Survey on Stereo Vision Disparity Map Algorithms

Effective detection of small targets plays a pivotal role in infrared (IR) search and track applications for modern military defense or attack Consequently, an effective small IR target detection algorithm based on a novel local contrast measure (NLCM) is proposed in this letter Initially, difference of Gaussian band-pass filter is employed to enhance target and suppress background clutter Then, a segmentation operation is implemented to obtain IR local regions of fixed size larger than general IR small target size Finally, the salient map is obtained using the NLCM, and an adaptive threshold is applied to extract the target region Experimental results on two real sequences show that the proposed method has better detection performance compared with conventional baseline methods

Effective Infrared Small Target Detection Utilizing a Novel Local Contrast Method

A Region Based Stereo Matching Algorithm Using Cooperative Optimization

Superpixel segmentation partitions an image into perceptually coherent segments of similar size, namely, superpixels. It is becoming a fundamental preprocessing step for various computer vision tasks because superpixels significantly reduce the number of inputs and provide a meaningful representation for feature extraction. We present a pixel-related Gaussian mixture model (GMM) to segment images into superpixels. GMM is a weighted sum of Gaussian functions, each one corresponding to a superpixel, to describe the density of each pixel represented by a random variable. Different from previously proposed GMMs, our weights are constant, and Gaussian functions in the sums are subsets of all the Gaussian functions, resulting in segments of similar size and an algorithm of linear complexity with respect to the number of pixels. In addition to the linear complexity, our algorithm is inherently parallel and allows fast execution on multi-core systems. During the expectation-maximization iterations of estimating the unknown parameters in the Gaussian functions, we impose two lower bounds to truncate the eigenvalues of the covariance matrices, which enables the proposed algorithm to control the regularity of superpixels. Experiments on a well-known segmentation dataset show that our method can efficiently produce superpixels that adhere to object boundaries better than the current state-of-the-art methods.

Superpixel Segmentation Using Gaussian Mixture Model

/pdf/uncertainty-estimation-for-stereo-matching-based-on-2ilppl35.pdf

Uncertainty estimation for stereo matching based on evidential deep learning

Cross-trees, edge and superpixel priors-based cost aggregation for stereo matching

The invention provides an ATM (automatic teller machine) intelligent monitoring system and an ATM intelligent monitoring method. The ATM intelligent monitoring system is used for performing video monitoring on an ATM or a user under the ATM monitoring scene, and consists of an image acquisition and processing module, a sensor and sensing data processing module, an intelligent analyzing module and an alarming module. The ATM intelligent monitoring method comprises various anomalous event detection methods, such as face detection and recognition, peep behavior detection, false advertisement and hangover detection, personnel wandering analysis, acute behavior alarming, yellow line crossing alarming, video signal abnormity monitoring, ATM bayonet and keyboard refit and image quality analysis. The ATM intelligent monitoring device is characterized in that the camera is firstly used for shooting the face of an ATM operator, an ATM keyboard, an ATM screen and the scene of a room where the ATM is located, the sensing data acquisition and processing module is used for acquiring vibration of the ATM and the signal of a position where the human body is located, and the intelligent analyzing module is used for analyzing human behavior and alarming for suspicious behavior. The problems that the traditional monitoring and analyzing system is large in workload, low in efficiency, easy in neglecting and failing to report the suspicious behavior, non-real-time alarming and the like are solved.

ATM (automatic teller machine) intelligent monitoring system and method

Stereo matching by using the global edge constraint

Learning to refine depth for robust stereo estimation

The invention relates to an intelligent detection system and method for video surveillance. The intelligent detection system for video surveillance is composed of a server-side and a client-side. The server-side is composed of a gateway server, a center management server, a task response server and a third-party server; the client-side is composed of an interactive interface and a network service client agent. The server-side manages the collection, processing, exchange and storage of data such as users and devices; the client-side provides the interactive interface for the users to achieve interaction between the users and the server-side. Aiming at the actual condition that existing video surveillance devices are scattered, the mode based on a network service platform is adopted to achieve the connection and data exchange between modules. Loads of a main server are relieved effectively through the distributed type deployment of the system. The number of video surveillance devices which are allowed to have access to the system is greatly increased, and the expandability of the system is high. When faults are detected, data calculation is carried out in a parallel manner, the calculation speed of the system is increased effectively, and the running and detection efficiency of the system is improved effectively.

Feiyang Cheng

Papers

Cross-trees, edge and superpixel priors-based cost aggregation for stereo matching

ATM (automatic teller machine) intelligent monitoring system and method

Stereo matching by using the global edge constraint

Learning to refine depth for robust stereo estimation

Intelligent detection system and method for video surveillance