Downsampling, oversampling, undersampling, subsampling, subsampling, and upsampling]

Downsample:2048hz is oversampling for the signal. In fact, as long as the signal is not overlapped (satisfying the nequest sampling theorem), The oversampling signal can be extracted, which is called "downsampling ". Sampling in the field is often limited by specific conditions, or there is no sampling rate of Hz, or debugging is very difficult. If R> 1, RFS/2 is much larger than the highest frequency FM of the audio signal, which makes the quantization noise mostly distributed in high-frequency areas outside the audio band, the quantitative noise distributed within the audio frequency band is reduced accordingly. Therefore, the signal-to-noise ratio of the system can be improved by filtering out the noise components above the FM through the low-pass filter. The original sampling frequency is 2048Hz, and the maximum allowable frequency of the signal is 1024Hz (satisfying the nequest sampling theorem). However, when the filter is used, the maximum frequency of the signal is 16Hz, in this case, the sampling frequency can be 32Hz (satisfying the nequest sampling theorem. The minimum value is 32Hz, which can be higher than 32Hz ). From 2048hz to 32Hz, one sample is taken every 64 samples. This reduces the sampling frequency by downsampling downsample ). The advantage of this is to reduce the data sample points, that is, to reduce the computing time, which is often used in real-time processing. Oversampling:Oversampling is defined as: sampling is performed at a frequency higher than that of nycept. The advantage is that the signal-to-noise ratio can be improved, but the disadvantage is that the processing of a large amount of data. Oversampling is used to sample the input signal at a frequency much higher than the nequest sampling frequency. The original sampling frequency of the digital audio system is FS, usually 44.1khz or 48 khz. If the sampling frequency is increased to r × FS, R is called the oversampling ratio and r> 1. In this digital signal, since the number of Quantizing BITs has not changed, the total quantization noise power remains unchanged. However, the spectral distribution of quantization noise changes, evenly distributed between 0 and 0 ~ The quantitative noise in the FS/2 band is scattered to 0 ~ RFS/2 frequency band. If R> 1, RFS/2 is much larger than the highest frequency FM of the audio signal, which makes the quantization noise mostly distributed in high-frequency areas outside the audio band, the quantitative noise distributed within the audio frequency band is reduced accordingly. Therefore, the signal-to-noise ratio of the system can be improved by filtering out the noise components above the FM through the low-pass filter. However, the effect of improving the signal-to-noise ratio by using this oversampling method is not obvious. Therefore, we must combine the noise shaping technology. Intuitively speaking, the signal after sampling is the continuation and superposition of the original signal in the frequency domain. The band-limited signal is usually the farther away from the center and the lower the amplitude. Therefore, the higher the sampling rate, the smaller the aliasing. the purpose of oversampling is to change the noise distribution, reduce the noise in the bandwidth of the useful signal, and then filter out the noise through the low-pass filter to achieve a better signal-to-noise ratio, it is generally used in sigma-deltadac or ADC. Oversampling: The noise can be extended to a higher frequency. After a low-pass filter is used, the SNR in the baseband can be improved. improve the time domain resolution to obtain better time domain waveforms; 2. improve the processing gain of the filter. When filtering in the frequency domain, the filter design becomes easier. 3. improves signal-to-noise ratio and collects waveform energy better when matching and filtering. 4. suppress the image to make it easier to increase the frequency of conversion and reduce the retention time requirements for the subsequent DA conversion. fractional Sampling timing is required. oversampling application: D/A conversion, but not necessarily oversampling. The oversampling technique is generally used at low speed (dozens of K to several m) high precision (such as 16bit 18bit .....). Da oversampling can be implemented using linear interpolation. Undersampling:When sampling frequency fs. max is greater than twice the maximum fmax frequency in the signal, that is, FS. max> = 2 FMAX, the digital signal after sampling completely retains the information in the original signal. Generally, the maximum frequency of the signal is 2.56-4 times. The sampling theorem is also called the nequest theorem. Under-sampling is a method that increases the bandwidth of the test device to achieve higher sampling frequency. According to the sampling theory, when sampling a complex signal (composed of several component signals with different frequencies), if the sampling clock frequency is less than twice the maximum frequency of the signal, then there will be a phenomenon called "alias. When the sampling clock frequency is low enough, it leads to an alias called "undersampling. A analog signal with a bandwidth of FB must have a sampling rate of FS> 2fb to avoid information loss. The minimum sampling frequency actually required is the function of the signal bandwidth, not only depends on its maximum frequency component. Generally, the sampling frequency must be at least twice the signal bandwidth, and the sampled signal cannot be
An integer multiple of FS/2 to prevent overlapping components. Undersampling is a very useful tool in Software Radio applications, but it must be very careful and careful to achieve good performance. Subsampling:Two sampling methods can be used for color TV images. One is to use the same sampling frequency to sample the brightness and chromatic aberration signals of the image, and the other is to sample the brightness and chromatic aberration signals at different frequencies. If the sampling frequency of the chromatic aberration signal is lower than that of the brightness signal, this sampling is called subsampling ). Subsampling is basically based on two features of a human's visual system. First, the human eye is less sensitive to the color signal than the Brightness Signal, by using this feature, we can remove some color-expressing signals from the image to make it invisible. Second, the human eye has a certain degree of ability to distinguish the image details, this feature can be used to remove high-frequency signals from the image, making it hard to detect. Subsampling uses this feature to compress color TV signals. Subsample:Definition: For a sample value sequence, sample several sample values at intervals. In this way, the new sequence is the subsample of the original sequence. The main reason for the change in sampling rate is that different modules of signal processing may have different sampling rate requirements. Compared with the initial continuous time signal, the subsample still needs to satisfy the sampling theorem. Otherwise, such subsample may cause signal mixing. Downsampling is the extraction method, which is one of the basic content in multi-rate signal processing. In different application scenarios, downsampling can bring many benefits. Take the most common digital receiver as an example. The sampling rate of the baseband signal to be obtained is equal to the symbol rate, which is relatively low, however, the general practice is not to directly sample the analog signal using this sampling rate, but to use a high (dozens or even hundreds of times) sampling rate, which can improve the signal-to-noise ratio of the sampled signal, then, the digital method is used to filter and extract signals at multiple levels until the sampling rate of the final signal is equal to the symbol rate. In this way, the signal-to-noise ratio gain obtained is the ratio of the initial sampling rate to the sampling rate of the final output signal. Before different sampling rates, there is a filter whose bandwidth corresponds to the sampling rate. The higher the sampling rate, the larger the bandwidth of the filter. For bandwidth noise (the noise bandwidth is higher than the highest sampling rate ), the passing noise power is higher (the noise power is the power spectrum density multiplied by the upper bandwidth, that is, the mean value of the square of the noise component in each sample value .) The signal power remains unchanged before and after sampling (the signal power is the square average of the signal component in each sample value ). For narrow band noise or narrow band interference (the noise or interference bandwidth is lower than the highest sampling rate), the following sampling method is used to obtain the conclusion that the SNR gain is the ratio of the initial sampling rate to the sampling rate of the final output signal. Or the signal-to-noise power is not much higher than the gain. Top Sampling:Sampling is a sample that collects analog signals. Sampling is a continuous signal of time and amplitude. Under the action of a sampling pulse, the signal is converted into a discrete signal of time and amplitude. Therefore, sampling is also called the waveform discretization process. The premise of the general nequest sampling theorem is that the frequency is limited by the band-limited Signal of (0, F. Sampling usually refers to downsampling, that is, extracting signals. In fact, both the upper sampling and lower sampling are performed on the digital signal. The sampling rate of the Secondary Sampling is compared with the sampling rate obtained from the original digital signal (for example, from the analog signal sampling, higher than the original signal is called the upper sampling, and smaller than is called the lower sampling. The essence of sampling is interpolation or interpolation. The upper sampling is the inverse process of the lower sampling, also known as upsampling or interpolating [1]. The application of incremental sampling in frequency division multiplexing is a good example. If these sequences were originally obtained by sampling continuous time signals at the nequest frequency, they must be sampled before being used in multiple channels. 1. oversampling is two times of the sampling frequency greater than the highest frequency (the nequest sampling rate). The actual low-pass signal sampling is about 2.5 times of oversampling. 2. undersampling is less than the nequest sampling rate, it should be a band-pass sampling. 3. The sampling rate is equivalent to the original digital signal (for example, from analog signal sampling) sampling Rate comparison, greater than the upper sampling, less than the lower sampling. The upper sampling and lower sampling are interpolation and extraction respectively. Http://blog.sina.com.cn/s/blog_6c8556dd01017r0q.html downsample the image (or subsampled) or downsample (downsampled) for two main purposes: 1. Make the image fit the size of the display area; 2. Generate a thumbnail of the corresponding image. The main purpose of an enlarged image (or upsampling) or image interpolation (interpolating) is to enlarge the original image so that it can be displayed on a display device with a higher resolution. The image scaling operation does not bring more information about the image, so the image quality will inevitably be affected. However, there are some scaling methods that can increase the image information so that the scaled image quality exceeds the original image quality. Http://baike.baidu.com/view/3038019.htm