Abstract EN:

Object visibility and detection are important issues, particularly for visual displays ergonomic guidelines. A better knowledge of the relationships betweeen displays physical factors and detection performance can contribute to improve quality and cost of the devices used to simulate tasks where visual factors are essential. After a review of the factors (perceptive and physical), the work suggests that target detection performance can be predicted using a photometric model based on the modulation transfer function (MTF) of the device. Two types of experiments have been designed to verify this hypothesis: the first one with a photo-simulation paradigm and the second with a video-simulation paradigm. In both cases two aspects are developped: a physical analysis of the visual device mtf characteristics and a study of the human detection responses. The results obtained in the photo-simulation experiment show that the distance at which the MTF curve stabilizes can be considered as a predictor of the liminar value, performed by subjects who are already skilled in target detection and acquisition. The reported data confirm the hypothesis except when targets are decoys. In this case, it is argued that their photometric signatures constitue sufficient cues that permit trained subjects to recognize and identify a decoy well before detecting the other targets. The mtf features of the video-simulation device have been measured using sinewave grating charts, composed of black and white bars with various contrasts, directions and spatial frequencies. The photometric data point out that the modulation transfer is much better in the horizontal direction than in the vertical direction. This inter-directional variation is also observed in the subjects' responses, notably in the range of the middle and high frequencies. Using such grating charts has led to propose a theorical model that permits to predict the distance at which a target can be detected. The subjects' responses confirm the predictions outcomed from this model. The experimental data show that mtf is a satisfactory reference to predict target detection performance. In the last part of the thesis, these data are discussed with regard to other models (Johnson criteria and signal detection theory). Finally, conclusions are drawn on the limits of a model based on MTF.