The signal-to-noise ratio (SNR) and noise equivalent radiance (NER) are important characteristics of an electro-optical systems sensor. A new method to improve the system’s SNR and reduce the NER was studied by analyzing the radiometric calibration data of the system. An image was obtained, when the system was illuminated by the uniform light source outputting from integrating sphere. The adjourn pixels were binned by n×n pixels (n=2, 3, 4, 5) in this image. And the digital numbers of binning pixels were added together. The new images were achieved. With 5×5 pixels binned, the SNR increased from 165.7 up to 842.3, which is improved almost 5 times. At the same time, the NER decreased from 0.077(W/m2/sr) to 0.015(W/m2/sr), which is declined by nearly 5 times. With the binning pixels up to n×n, the SNR and NER were enhanced and decreased almost by n times, respectively.