Human trapped in a parked car recognition using thermal image approach

This study proposes thermal imaging technique to enhance visualization of a detected human face inside a detected car. Although an infrared camera it is more commonly used in industrial and military, but software application has the advantages by image processing an original image into a thermal imaging representation that only used a conventional CCTV camera in public areas. The algorithms started with car detection, counting and tracking as a primary step in a car park, followed by detecting a human face presence inside a car and finally implementing thermal image processing to further ensure the real situation happened correctly before issuing an error warning. Based on seven unbiased scenarios experiments, the accuracy of a standalone thermal imaging identification is higher than standalone human face detection for recognizing the human faces by 4.02%. Hence, a combined method improves the performance of thermal image identification for detecting human face inside a car in terms of accuracy by 2.97% as compared with the conventional human face recognition technique in the situations.