The currently used single-channel algorithms for land surface temperature (LST) retrieval are mainly proposed for sensors with limited scopes. The coverage of the infrared camera (IRS) onboard HJ-1B satellite is broad, and the view zenith angle of IRS channel 4 can reach ±33°. Therefore, the effects of view angle should be eliminated when retrieving LST with IRS4. Based on atmospheric radiative transfer simulation, this research explored a look-up table for view zenith angle (VZA)-coefficients of atmospheric functions and then proposed an improved single-channel algorithm for estimating LST from IRS4 images. According to the fact that the overpass time of HJ-1B and Terra are very close, the research investigated the possibility to integrate the water vapor provided by MODIS into the improved single-channel algorithm. In addition, this research proposed a practical routine to calculate the land surface emissivity of IRS4. Validation shows that the average absolute error of the improved algorithm is below 1.1K, which is about 0.1~0.7K lower than the algorithm without considering the view zenith angle. Application to the HJ-1B satellite imagery indicated that land surface temperatures retrieved with the improved algorithm are consistent with those provided by MODIS product.