Abstract: Based on MATLAB language, the paper compares with several binarization algorithm: Ostu algorithm, the algorithm for threshold selection based on arithmetic mean of gray value and the adaptive threshold selection which combines the global threshold with local threshold. The indexed ECG images are converted to RGB ones. And then binarization processing is implemented to the R-array, G-array and B-array separately by using the methods above. "OR" operation is experimented with the gotten arrays so as to get the result. The experimental results show that binarization approaches based on global threshold discussed in theis paper are useful and can remove the background noise, background grid and the marks at QRS complexes and keep the useful ECG signals.