JPS6314290B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a spectrophotometer with improved signal-to-noise ratio.

Generally, the S/N ratio is improved by applying an averaging process called smoothing to the measurement results by multiplying them by a value weight. Simply put, the fine irregularities in the measurement results are averaged out to form a smooth curve. Therefore, there is a drawback that the measurement results are distorted and some of the true value changes that the object to be measured have are also lost. In the case of a spectrophotometer, the spectral resolution deteriorates.

The present invention was made in order to improve the above-mentioned drawbacks in the smoothing operation, and the distribution shape of the value weight used in the averaging process is changed in conjunction with the slit width of the spectrophotometer, thereby minimizing the decrease in resolution. Finally, it made it possible to improve the S/N ratio. The wavelength resolution of the spectrum obtained by a spectrophotometer is determined by the slit width. The spatial frequency distribution of the spectral figure obtained by the spectrophotometer does not have frequency components higher than a certain frequency determined by the slit width. Therefore, high frequency components appearing in actual measurement results are noise. Therefore, from the raw data obtained by the spectrophotometer, we use a functional filter that removes components with a frequency higher than the highest frequency component determined by the slit width among the spatial frequency components determined by the slit width. Spatial frequency components were removed. In this case, the characteristics of the functional filter are changed in conjunction with the slit width. The present invention will be explained below with reference to Examples.

Figure 1 shows an idealized spectroscopic spectrum, hypothetically s
(λ). Figure 2 shows the optical device function of a spectrophotometer, which is the output when a single wavelength of light is put into the spectrophotometer and wavelength scanned, and is modulated so that the area under the curve is 1. Yes, temporarily expressed as (λ). If this is an ideal spectrophotometer, this would be a →
It is 0. Actually, it has a certain width as shown in the figure, and its width 2a is related to the slit width, and the wider the slit width is, the wider it becomes, and the height of the curve becomes lower. Figure 3 is a spectral image obtained by a spectrophotometer with the instrument function shown in Figure 2. When expressed as x (λ), x (λ) = ∫∝ _- ∝s (u) (u - λ )du ……(1) is the relationship. When both sides of equation ₍ 1) are Fourier transformed, X ⁽ ω)=S(ω)・F(ω)...(2) However ^, It can be written in the form of du. Since the Fourier transform shows the frequency distribution by considering the original function as the sum of sine functions with various frequencies, X(ω) shows the spatial frequency distribution of the spectral image in Figure 3, This is shown in FIG. According to Figure 4, X(ω)
It can be seen that |ω| > 2π/a, which means that it is almost 0 and does not contain any frequency components higher than ω=2π/a. Therefore, even if a low-pass filter having the characteristics shown by B(ω) in FIG. ) does not change its shape. Since B(ω) is a Fourier transform of a certain function, when the original function is determined, it becomes b(λ) in FIG. When the output x(λ) of a spectrophotometer is input to a device having the device function shown in Figure 5, the shape of the output hardly changes; If noise larger than /a is mixed in, such noise will be removed. Mathematically, it is a transformation of x(λ) as follows: ∫∝ _- ∝x(u)b(λ−u)du...(3) Specifically, as shown in Figure 6, To obtain the final measured value at the wavelength λ, take the value of x(λ) at many points in a range of about a/2 before and after λ, and multiply it by the corresponding value of b(u). It is to seek the sum.

In the present invention, the photoelectric conversion output was simply amplified and recorded in the conventional spectrophotometer, but it is further passed through a circuit that calculates the above equation (3) before being recorded. The function b(λ) in is made variable in conjunction with the setting of the slit width of the spectrophotometer. FIG. 7 shows the configuration of the apparatus of the present invention.

1 is a light source, M is a spectrometer, 2, 2 is an entrance and exit slit, 3 is a slit width detector, 4 is a sample, 5
is a photoelectric converter, 6 is an amplifier, and 7 is a weighted average circuit that performs the calculation of equation (3) above.

The weighted average circuit 7 samples the output of the amplifier 6 at suitably set wavelength intervals in conjunction with the wavelength scanning of the spectrometer M, converts it from analog to digital, and stores the sampled output in the first memory b(u). A second memory stores values of u for each unit of u, and a first memory stores sampling values (simply referred to as sampling values) of the outputs of n amplifiers 6 before and after the specified wavelength λ. The corresponding b is read out from the second memory in order from
It consists of a sequence control device that reads out the value of (u), multiplies the two, cumulatively adds the multiplication results, and displays or records all the addition results. Here, the shape of b(u) can be changed in conjunction with the slit width setting as follows. b(u) is a symmetrical function as shown in Figure 5, and the point a/2 changes in conjunction with the setting of the slit width, and as the slit width increases, the width of the peak increases in proportion to it and is inversely proportional to the width of the slit. The height of the mountains decreases. Therefore, the slit width is divided into appropriate units from the minimum to the maximum, and when the width is within the minimum division, the memory of b(u) is read every p units, and the width becomes wider by one division. b for each unit subtracted by one from p according to
The memory of (u) is read out, and the value obtained by multiplying the other read value by p/(p-k) is used for the weighted average calculation. On the other hand, as the slit width becomes wider, the sampling values read from the first memory are also read from a wider range. In this way, the wider the slit width, the more data will be used for the weighted average calculation. The fact that the slit width is wide means that the resolution is set that low, so calculations should be easy, so based on the above relationship,
The data read from the memory and the data read from the second memory are related to each other, and in actual calculations, it is sufficient to thin out the data appropriately, and the law for thinning is arbitrary. , these are specified by a program given to the control circuit.

As mentioned at the beginning, in a spectrophotometer, weighted average processing is already applied to the true spectral image in a purely optical manner in relation to the slit width, and spatial frequency components of 2π/a or higher are cut out. Therefore, the spatial frequency in the actually obtained spectral recording
When a fluctuation component of 2π/a or more is included,
This is noise caused by causes other than purely optical, such as fluctuations in the light source, noise in electrical circuits, etc. The present invention removes such noise at the final stage of the device, and does not modify the spatial frequency data that the spectrophotometer allows to be transmitted purely optically, resulting in less distortion and a good S/N ratio. Spectral data can be obtained.

[Brief explanation of the drawing]

Figure 1 is a graph of an idealized spectral image,
Fig. 2 is a graph of the optical device function of the spectrophotometer, Fig. 3 is a graph of the spectral image actually obtained, Fig. 4 is a graph showing the Fourier transform of Fig. 3, and Fig. 5 is a graph of the optical device function of the spectrophotometer. FIG. 6 is a graph showing the characteristics of the spatial frequency filter used, FIG. 6 is a diagram explaining the weighted average processing operation, and FIG. 7 is a block diagram showing the configuration of the apparatus of the present invention. 1...Light source, M...Spectrometer, 2...Slit,
3... Slit width detector, 4... Sample, 5... Photoelectric converter, 6... Amplifier, 7... Weighted average circuit.

Claims (1)

[Claims] 1. Means for generating a value weight function in conjunction with the setting of the slit width so that when the slit width is a, the Fourier transformed form forms a rectangle with both ends at ±2π/a, and a value weight function read from this device. A spectrophotometer equipped with a weighted averaging circuit that performs weighted averaging processing on the final output that has been photoelectrically converted.