Low Resolution Person Detection with a Moving Thermal Infrared Camera by Hot Spot Classification

Abstract

In many visual surveillance applications the task of person detection and localization can be solved easier by using thermal long-wave infrared (LWIR) cameras which are less affected by changing illumination or background texture than visual-optical cameras. Especially in outdoor scenes where usually only few hot spots appear in thermal infrared imagery, humans can be detected more reliably due to their prominent infrared signature. We propose a two-stage person recognition approach for LWIR images: (1) the application of Maximally Stable Extremal Regions (MSER) to detect hot spots instead of background subtraction or sliding window and (2) the verification of the detected hot spots using a Discrete Cosine Transform (DCT) based descriptor and a modified Random Naïve Bayes (RNB) classifier. The main contributions are the novel modified RNB classifier and the generality of our method. We achieve high detection rates for several different LWIR datasets with low resolution videos in real-time. While many papers in this topic are dealing with strong constraints such as considering only one dataset, assuming a stationary camera, or detecting only moving persons, we aim at avoiding such constraints to make our approach applicable with moving platforms such as Unmanned Ground Vehicles (UGV).

Linear Support Tensor Machine With LSK Channels: Pedestrian Detection in Thermal Infrared Images

“…Following the evaluation technique established for detecting pedestrians in the visible spectrum [3], we report here the miss rate versus FPPI graph as a measure of detector performance. This is in contrast to earlier notion of false positive per window (FPPW) as used to evaluate the pedestrian detector [1], [7]. Miss rate versus FPPI is also in contrast to the precisionrecall curves that are more traditionally followed in other areas Fig.…”

“…Infrared and thermal imaging sensors, which provide excellent visible cues in unconventional settings (e.g., night time visibility), have historically found their use limited to military, security and medical applications. However, with increasing image quality and decreasing price and size, some of the thermal sensing devices are finding commercial deployment for home and office monitoring as well as automotive applications [6], [7]. Research effort has so far been limited in this domain for building reliable and efficient computer vision systems for infrared thermal image sensors.…”

Vehicle classification based on images from visible light and thermal cameras

“…In addition, a thermal camera with UAV was used to detect objects in the ocean surface [24]. In [25,26], people were detected and classified by a moving thermal infrared camera with the lowresolution images. In this paper, we propose a smart surveillance system to detect and classify vehicles during the day and at night.…”

Classifying objects in LWIR imagery via CNNs

“…When we consider the dearth of publicly available, large scale datasets containing IR imagery, the sparsity of such methods may be explained by the fact CNNs are still recent developments in computer vision, as there are many more classification schemes for thermal band data using older machine learning tools and practices. 1,13,14 While these works do show impressive performance for IR image classification tasks over a small number of objects, they will be outperformed by deep learning methodologies that enable more descriptive features to be learned. 12 The automatic discovery of better features leads to increasingly effective object recognition performance.…”