In this paper, we propose a novel image segmentation algorithm that is based on the probability distributions of the object and background. It uses the variational level sets formulation with a novel region based term in addition to the edge-based term giving a complementary functional, that can potentially result in a robust segmentation of the images. The main theme of the method is that in most of the medical imaging scenarios, the objects are characterized by some typical characteristics such a color, texture, etc. Consequently, an image can be modeled as a Gaussian mixture of distributions corresponding to the object and background. During the procedure of curve evolution, a novel term is incorporated in the segmentation framework which is based on the maximization of the distance between the GMM corresponding to the object and background. The maximization of this distance using differential calculus potentially leads to the desired segmentation results. The proposed method has been used for segmenting images from three distinct imaging modalities i.e. magnetic resonance imaging (MRI), dermoscopy and chromoendoscopy. Experiments show the effectiveness of the proposed method giving better qualitative and quantitative results when compared with the current state-of-the-art.

A method of detecting target objects in still images despite any kind of geometrical distortion is demonstrated. Two existing techniques are combined, each one capable of creating invariance to various types of distortion of the target object. A maximum average correlation height (MACH) filter is used to create invariance to orientation and gives good tolerance to background clutter and noise. A $logr-θ$ mapping is employed to give invariance to in-plane rotation and scale by transforming rotation and scale variations of the target object into vertical and horizontal shifts. The MACH filter is trained on the $logr-θ$ map of the target for a range of orientations and applied sequentially over regions of interest in the input image. Areas producing a strong correlation response can then be used to determine the position, in-plane rotation, and scale of the target objects in the scene.

Many Multi-Object Tracking (MOT) approaches exploit motion information to associate all the detected objects across frames. However, traditional tracking-by-detection (TBD) methods, relying on the Kalman Filter, often work well in linear motion scenarios but struggle to accurately predict the locations of objects undergoing complex and non-linear movements. To overcome these limitations, we propose ETTrack, a novel motion prediction method with an enhanced temporal motion predictor. Specifically, the motion predictor integrates a transformer model and a Temporal Convolutional Network (TCN) to capture both long-term and short-term motion patterns, and it predicts the future motion of individual objects based on the historical motion information. Additionally, we propose a novel Momentum Correction Loss function that provides additional information regarding the motion direction of objects during training. This allows the motion predictor to rapidly adapt to sudden motion variations and more accurately predict future motion. Our experimental results demonstrate that ETTrack achieves a competitive performance compared with state-of-the-art trackers on DanceTrack and SportsMOT, scoring 56.4 $$\%$$% and 74.4 $$\%$$% in HOTA metrics, respectively. Our work provides a robust solution for MOT in complex dynamic environments, which enhances the non-linear motion prediction capabilities of tracking algorithms.

A hybrid optical/digital configuration to implement a highspeed correlator is described. The input data is digitally Fourier transformed at video rates utilizing specialized digital signal processing chip sets and the computed spectral phase data passed to an electrically addressed phase modulating spatial light modulator (SLM). Successive templates, which are stored as angle multiplexed volume holograms in a photorefractive crystal, are reconstructed and optically correlated with the current input scene during an interval in which the SLM display is held stable. In this way, shift invariance, one of the major advantages of the correlation technique, is preserved. Of critical importance to the speed of operation are the energy losses of the coherent wavefront through the system and the rate at which correlation plane data can be converted to an electronic signal; both of these issues are considered to estimate that at least a 3-kHz search rate is achievable from a 1000 template memory with system components currently available.

There are an increasing number of high-occupancy-vehicle lanes throughout the world. In these lanes each vehicle is required to have two or more occupants. These laws are currently manually enforced. This paper presents a new system for automating the enforcement process, using color image segmentation techniques to find the windshield area, and face detection to count the number of occupants within a vehicle. © 2004 Society of Photo-Optical Instrumentation Engineers.