The robotics infrared sensor skin is a new outward sensing system which has the characteristics of abundant information, high performance on real-time processing and non-contact measurement for its environment. The infrared sensor skin is widely used in the multi-joint robot system for real-time obstacle avoidance. However, the output of mini-type infrared sensors in sensor skin is subjected to noise disturbances caused by visible light, electrical source with its work frequency and fluorescent lamp etc. So, it is very difficult for the traditional analog circuit to meet the system's demand of real-time and veracity. In order to improve the systemic performance, a kind of all-digital signal processing technique based on the digital signal process (DSP) is presented. Signal processing is implemented by using fast Fourier transform (FFT) method. By analyzing the error caused by the FFT method, error compensation is realized by using Hanning window. The results of simulations and experiments show that this technique can reduce the effect of noises and provide accurate and real-time environmental information by using the infrared sensor skin.