The fingerprint region of most gases is within 3 to 14μm. A mid-wave or long-wave infrared thermal imager is therefore
commonly applied in gas detection. With further influence of low gas concentration and heterogeneity of infrared focal
plane arrays, the image has numerous drawbacks. These include loud noise, weak gas signal, gridding, and dead points,
all of which are particularly evident in sequential images. In order to solve these problems, we take into account the
characteristics of the leaking gas image and propose an enhancement method based on adaptive time-domain filtering
with morphology. The adaptive time-domain filtering which operates on time sequence images is a hybrid method
combining the recursive filtering and mean filtering. It segments gas and background according to a selected threshold;
removes speckle noise according to the median; and removes background domain using weighted difference image. The
morphology method can not only dilate the gas region along the direction of gas diffusion to greatly enhance the
visibility of the leakage area, but also effectively remove the noise, and smooth the contour. Finally, the false color is
added to the gas domain. Results show that the gas infrared region is effectively enhanced.
Leakage of dangerous gases will not only pollute the environment, but also seriously threat public safety. Thermal infrared
imaging has been proved to be an efficient method to qualitatively detect the gas leakage. But some problems are remained,
especially when monitoring the leakage in a passive way. For example, the signal is weak and the edge of gas cloud in the
infrared image is not obvious enough. However, we notice some important characteristics of the gas plume and therefore
propose a gas cloud infrared image enhancement method based on anisotropic diffusion. As the gas plume presents a large
gas cloud in the image and the gray value is even inside the cloud, strong forward diffusion will be used to reduce the noise
and to expand the range of the gas cloud. Frames subtraction and K-means cluttering pop out the gas cloud area.
Forward-and-Backward diffusion is to protect background details. Additionally, the best iteration times and the time step
parameters are researched. Results show that the gas cloud can be marked correctly and enhanced by black or false color,
and so potentially increase the possibility of gas leakage detection.
Standoff detection of gas leakage is a fundamental need in petrochemical and power industries. The passive gas imaging
system using thermal imager has been proven to be efficient to visualize leaking gas which is not visible to the naked
eye. The detection probability of gas leakage is the basis for designing a gas imaging system. Supposing the performance
parameters of the thermal imager are known, the detectivity based on electromagnetic radiation transfer model to image
gas leakage is analyzed. This model takes into consideration a physical analysis of the gas plume spread in the
atmosphere-the interaction processes between the gas and its surrounding environment, the temperature of the gas and
the background, the background surface emissivity, and also gas concentration, etc. Under a certain environmental
conditions, through calculating the radiation reaching to the detector from the camera's optical field of view, we obtain
an entity "Gas Equivalent Blackbody Temperature Difference (GEBTD)" which is the radiation difference between the
on-plume and off-plume regions. Comparing the GEBTD with the Noise Equivalent Temperature Difference (NETD) of
the thermal imager, we can know whether the system can image the gas leakage. At last, an example of detecting CO<sub>2</sub> gas by JADE MWIR thermal imager with a narrow band-pass filter is presented.