The amplitude of the instantaneous value obtained by sampling is discrete, that is, a set of specified levels is used to represent the instantaneous sampling value with the closest level value.
Or it refers to dividing the continuous range of input signal amplitude into finite non-overlapping subintervals (quantization levels), and each subinterval is represented by a certain value in the interval, and the input signal falling in it will be output by this value, thus changing the continuous input signal into an approximate signal of finite discrete value levels.
Extended data:
Whether it is uniform quantization of continuous gray values sampled at equal intervals and stratified sampling or non-uniform quantization at unequal intervals, all gray values between two quantization levels (called two decision levels) are represented by one quantization value (called quantization level output by quantizer).
The sampled images are only spatially dispersed into pixel (sample) arrays. However, the gray value of each sample is still a continuous variable with infinite values, which must be converted into finite discrete values and given different codewords to truly become a digital image.