JPH0773197B2 - Floating binarization method for image processing - Google Patents

Description

TECHNICAL FIELD The present invention relates to a floating binarization method for image processing for binarizing an image signal obtained from an image pickup device such as a TV camera with high accuracy.

(Background Art) When a character pattern, a wiring pattern, or the like as shown in FIG. 3 (a) is imaged by an image pickup device such as a TV camera, ideally, a video signal such as (b) (video for one scan) Signal is shown).

However, in general, noise that humans can unknowingly exclude and recognize only meaningful signals such as character patterns, for example, unevenness of lighting or distortion that is a gradual change due to location or characteristics of a camera , Even if a uniform image is picked up, a signal waveform including distortion is obtained as shown in (c) due to the effect of shading characteristics such as the output signal changing from left to right on the screen gradually (smallly). It is often done.

By the way, as a conventional image processing technique, an image signal of each point of a monochrome image is divided into 1 or 0 by a certain threshold value, and an object such as an object to be recognized is simply converted into an image signal of brightness of each point. A binarization method or the like is known as a method of compressing information such that it is 1 if it is larger than the threshold value and 0 if it is smaller than the threshold value.

However, in the conventional binarization method or the like, even if the binarization is performed by a fixed binarization threshold, or even if the floating binarization threshold is adopted, it does not sufficiently follow the original signal. Since the binarization accuracy is deteriorated due to the change of the frequency component of the signal of the edge portion between the background and the background, and the above-mentioned illumination unevenness and shading are easily affected, the object such as an object to be recognized is sufficient. There were drawbacks such as not being able to recognize it.

(Object of the Invention) The present invention has been proposed in view of the above points, and an object of the present invention is to provide a video signal including the above distortion obtained when a character pattern, a wiring pattern, or the like is imaged. It is an object of the present invention to provide a floating binarization method for image processing that can accurately recognize an object by binarizing only a necessary signal part with high precision.

(Disclosure of the Invention) Hereinafter, the present invention will be described in detail with reference to the drawings.

FIG. 1 shows a signal waveform which undergoes processing and is transformed by the floating binarization method for image processing of the present invention. The principle thereof is an image obtained from an image pickup device such as a TV camera. The original signal a, which is a signal, is differentiated, and only the part with a large change having a change amount of a certain value or more is taken out, and the time t ₁ at the center thereof is calculated to know the central time of the change portion,
The signal before and after this time is checked to calculate the intermediate level (weighted average, simply arithmetic average) of the two signals before and after, and is used as the threshold value for that section to binarize the image signal. .

Hereinafter, a procedure for obtaining the floating binarization threshold value b that appropriately binarizes the original signal a of the video signal obtained by imaging the character pattern, the wiring pattern, and the like with an imaging device such as a TV camera will be described.

First, the obtained original signal a is first-order differentiated to calculate the absolute value of the differential value. At this time, the differential signal c is obtained. The convex portion of the waveform of the differential signal c corresponds to the edge portion of the waveform of the original signal a. In order to extract this edge portion, threshold processing is performed on the differential signal c with an appropriate differential threshold d. At this time, the threshold value d is appropriately determined according to how much a slow change is ignored, that is, what kind of property the input image signal includes an unnecessary signal, that is, the above distortion. As a result, a differential binarized signal e is obtained. The center of the high level period of the differential binarized signal e is extracted, and this time is defined as t ₁ . Original signal a at time t ₁
Corresponds to the central part of the edge portion of.

This will be described below with reference to FIG. Now, let the original signal a be the function y
= F (t). At this time, the time t _{1 of the} original signal a
Write point, i.e. each value of the point shifted by Delta] t before and after from the notice point _{t 1 f (t 1 + Δt} ), and f (t ₁ -Δt) in. If the point obtained by dividing this binary value into m: n is g (t ₁ ), Becomes This division of m: n is related to averaging after weighting the previous signal in time, or averaging after weighting the later signal, depending on what kind of waveform the waveform corresponds to. Can be decided

Similarly, g (t ₀ ), g (t ₂ ) is also obtained for times t ₀ and t ₂ ,
Point [t ₀ , g (t ₀ )] and point [t ₁ , g (t ₁ )], point [t ₁ , g
The straight lines connecting (t ₁ )] and the points [t ₂ , g (t ₂ )] are set as floating binarization thresholds. That is, the floating binarization threshold value b is in the interval t ₀ ≤t <t ₁ . In the interval t ₁ ≤t <t ₂ And, in general, in the interval t _n-1 ≤t <t _n Can be written.

Then, the floating binarized signal f is obtained by thresholding the original signal a with this floating binarized threshold b (= y).

Here, the original signal a, which is a video signal obtained by imaging a character pattern, a wiring pattern, or the like with an imaging device such as a TV camera, is expressed as a function y = f (t) that changes with time t
Actually, the input original signal a is sampled at a predetermined time interval, D / A is changed, and then g (t ₁ ) at each time for the original signal a input by the computer
Alternatively, g (t ₀ ), g (t ₂ ) and the like are obtained using the above calculation means to obtain the floating binarized signal f. The time t of the original signal a
Alternatively, the point deviated by Δt before and after the time t is appropriately determined depending on the accuracy of the signal to be obtained and the device used.

(Effects of the Invention) As described above, in the present invention, when binarizing a character pattern, a wiring pattern, or the like, a floating binarization threshold value that follows the waveform of the original signal is set. Therefore, there is an effect that binarization can be performed with high accuracy and an object can be recognized with high accuracy without being affected by distortion such as uneven illumination and shading.

[Brief description of drawings]

FIG. 1 is a diagram showing a signal waveform which is processed and deformed by the floating binarizing method and apparatus of the present invention, and FIG.
FIG. 3 is a diagram showing details of the threshold value, and FIG. 3 is a diagram showing a captured image and a video signal. a ... original signal, b ... floating binary threshold c ... differential signal, d ... differential threshold e ... differential binary signal, f ... binary signal

Claims (1)

[Claims] 1. An original signal is a video signal obtained by capturing an image of a character pattern, a wiring pattern or the like with an image pickup device such as a TV camera, and the original signal is first differentiated to calculate a differential signal indicating the absolute value of the differential value. A step of thresholding the differential signal with an appropriate differential threshold value to obtain a differential binary signal, and a step of calculating a central time t _n of a high level period of the differential binary signal. When, determining a value of the original signal at the calculated time t time shifted by Δt before and after each than _n, the 2 value of the m: values were divided into n g
The step of obtaining (t _n ) and the point [t _n-1 , g (t _n-1 )] and the point [t _n , g (t _n )] in the interval of t _n-1 ≤t <t _n Straight line And a binarization of the original signal by using the straight line y as a floating binarization threshold value.