JPS6352367B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention relates to a same-scene frame determination method for determining whether consecutive frames in a negative film are the same scene during photographic printing. A single negative film often contains frames of the same scene. It is desirable to print the same scene at the same density. Based on this recognition, a preliminary inspection station has been proposed that simultaneously observes a long series of original images and inputs the same correction value at high speed if the original images of the same scene are consecutively lined up. (For example, Tokukai Sho
No. 47-6932) Most of the same scene usually differs in details depending on the camera position, angle, size of the main subject, etc. Therefore, in a method of scanning the screen and analyzing photometric data to determine the exposure amount for printing, as shown in Japanese Patent Application Laid-Open No. 52-23936, characteristic values for exposure control, such as maximum density Dmax, minimum density, etc. There is a big difference in Dmin. Furthermore, even if the difference between Dmax and Dmin is small, a difference occurs when the calculation output is converted into a stepwise density key. As a result, even the same scene is printed with different densities. Incidentally, the same scene is often shot consecutively. Therefore, the detected values of the total area average transmitted density LATD and shooting light quality of several frames before and after the frame to be measured are compared with the detected values of the frame, and if the difference is within a certain value, it is determined that these are the same scene. A determination method is known (Japanese Unexamined Patent Publication No. 49-40942). In this way, when it is determined that the scenes are the same, the same scene frames are printed using the intermediate value of the exposure amounts of these same scenes so that they have the same density. It is known that when all the transmitted light of a color negative is integrated and mixed, it becomes gray or a constant hue close to gray (embassy theory). Therefore, daylight,
For most of the strobe lights, the detected values of the shooting light quality are also similar. Furthermore, even if the LATDs are similar, for example, the LATDs of a standard scene and a strobe photography scene may be similar, and in this case, it is necessary to give different exposure amounts. Determining the similarity of frames using only the detected values of LATD and photographing light quality involves the problem that even different scenes with different scene configurations are determined to be the same scene. Furthermore, with a method that only compares LATD, even if the scene is the same, if the exposure level is changed during shooting, it will not be determined as the same scene, and even if the same scene is backlit and the area of the sky is slightly different, there will be a difference in LATD. , the scenes will not be determined to be the same scene. As described above, the conventional method has drawbacks in that it does not include the composition of the scene, and that it is determined that they are similar only when the LATD and photographing light quality detection values for each color are all approximate. Furthermore, with the conventional method, it is necessary to memorize all the comparison data for several consecutive frames, the relevant frame and the photometry frame are different and require complicated control, and it is necessary to examine several frames before and after each frame. However, it has drawbacks such as being inefficient. SUMMARY OF THE INVENTION In view of the above-mentioned problems, it is an object of the present invention to provide a method for determining same-scene frames that improves the accuracy of determining same-scene frames. The determination method of the present invention is characterized in that the determination is made using only the photometric data of the current frame and the previous frame, and by repeating this process, it is possible to determine a series of identical scenes with high accuracy. For this determination, the present invention utilizes characteristic values representing the difference in color between the current frame and the previous frame, the difference in scene structure (picture pattern), and the like. Characteristic values that represent color differences include ΔRG, which represents the difference between R (red) and G (green), and G and B.
(Blue) sum of ΔGB representing the difference | ΔRG | + | ΔGB
| is used. These are RG(=LATD(R)−
It represents the color difference between two frames on the color coordinates consisting of the LATD(G)) axis and the GB (=LATD(G)-LATD(B)) axis. As the characteristic value representing the difference in the structure of the scene, a combination of the average density of each section divided as shown in FIG. 2 is used. i.e. top/bottom,
The density distribution in the screen can be compared based on the density difference between right/left and center/periphery, and it can be determined, for example, whether there is a high density distribution like the sky at the top of the screen. Furthermore, it is also possible to make a determination by adding differences in histogram information and differences in contrast information to the above characteristic values. As this histogram information, use (Dmin+Dmax)/2-LATD(W) (where Dmin is the lowest density of neutral gray, Dmax is the highest density, and LATD(W) is the total area average transmission density of neutral gray). When this value is large, it indicates that there are many low-density parts in the screen, that is, the scene has many shadows, and when this value is small, it indicates that the scene has many highlights. As the contrast information, Σ|D′i+1−D′i|/n can be used. This represents the average contrast of the entire screen, and is used to determine whether a scene has low contrast, such as a landscape, or a scene with high contrast, such as a winter green scene. In the present invention, the following equation, which is a combination of the above information differences between two consecutive frames, can be used as a determination equation for the same scene. X=K ₁ +K ₂ (|ΔRG|+|ΔGB|)+K ₃ (|ΔUL|+|
ΔRL|)+K ₄ |ΔCP|+K ₅ |ΔDB| Note that K ₁ to K ₅ are constants including 0 here. According to this method, even if adjacent frames have different exposure amounts at the time of shooting, in the case of the same scene, the difference in scene composition (picture pattern) and color difference are used instead of including negative density information. By the way,
It is possible to correctly determine that the frames are the same scene.
Also, even if there is a difference in LATD due to a slight difference in the area of the sky in the same scene,
For example, even if |ΔUL| shows a slightly large value, since the difference in other characteristic values for determination is small, it is possible to determine that the scenes are the same scene as a comprehensive result. The most important thing in determining identical scene frames is that it is not only necessary to determine exactly the same picture as the same scene, but also to determine that slightly different pictures are the same scene. For example, 1 at the same location
It is necessary to determine that the scene is the same in the following cases: a frame with a famous person in it and a frame with three people in it, a frame with the same location but with slightly different camera angles, and a frame with a person in the center and a frame with a person off to the right when shooting with a strobe light. be. The exact same scene will be output with the same exposure compensation amount shown in JP-A No. 52-23936, but for frames that are slightly different, the output results may be printed too darkly or too lightly even though they are within the allowable range. Prints may appear on the eyes, and variations in density become noticeable. The method of the present invention allows the same scene frames to be printed at the same density even in such a case. The configuration of the present invention will be explained in detail below. FIG. 1 shows an example of a negative film. Frames 1 to 3 are the same scene. These same scenes are often photographed consecutively. Therefore, the characteristic value obtained by measuring frame 1 and the exposure correction value calculated using the exposure control formula in JP-A No. 52-23936, for example, are stored, and the judgment formula is calculated together with the characteristic value of frame 2. input to the calculation section. If it is determined that they are the same scene, the exposure correction value for frame 2 is added to the exposure correction value for frame 1, and the next frames 3 and 2 are judged. If this frame 3 is also determined to be the same scene, the exposure correction value of frame 3 is added to the exposure correction value added for the previous frame, and then frames 4 and 3 are
is determined. When it is determined that this frame 4 is a different scene (discontinuous) from frame 3, the exposure amount correction values for frames 1 to 3 are output. This exposure amount correction value is an average value obtained by dividing the exposure amount correction value that was added when the frames were determined to be continuous frames by the number of frames. Frames 1 to 3 are printed using this average value and are finished with the same density. When frame 4 is determined to be a different scene, the characteristic value and exposure correction value of frame 4 used in the determination formula are stored, and it is determined that frame 4 is the same scene as frame 5. If it is determined that the scene is a different scene, the stored exposure correction value is output. The characteristic value and exposure correction value of frame 5 are then stored and prepared for determination as the next frame. The following equation is used to determine the same scene. X=K ₁ +K ₂ {|ΔRG|+|ΔGB|}+K ₃ {|ΔUL|+
| _ΔRL |｝＋K ₄ | _ΔCP |+K ₅ |ΔDB|
R) − LATD(G)) | |ΔGB|=|(LATD′(G)−LATD′(B))−(LATD(G)−L
ATD(B)) | |ΔUL|=|(DL′−DU′)−(DL−DU)|||ΔRL|=|(DRI′−DLF′)−(DRI−DLF)| |ΔCP|=|( (Dmin′+Dmax′) 1/2−LATD′(W
)) − ((Dmin+Dmax)1/2−LATD(W)) | |ΔDB| = Σ(D′i+1−D′i)/n−Σ(Di+1−Di
)/n (This is the average value of the density difference between adjacent measurement points, and by determining it for the measurement points on the entire screen, it represents the average contrast of the entire screen.) LATD (W): Neutral gray Total area average transmission density LATD(R): Red total area average transmission density LATD(G): Green total area average transmission density LATD(B): Blue total area average transmission density Dmin: Neutral gray minimum density Dmax : Maximum density of neutral gray DC: Average density at the center of the screen DF: Average density at the periphery of the screen DU: Average density at the top of the screen DL: Average density at the bottom of the screen DRI: Average density at the right side of the screen DLF: Average density at the left side of the screen It is concentration. Further, those characteristic values with a dash (') attached thereto are the characteristic values of the previous frame, and those without any attachment are the current frame whose scene is judged as to whether or not it is the same scene as the previous frame. If the determination value in the determination formula is equal to or greater than a certain value α, it is determined that the previous frame and the current frame are the same scene. That is, X≧α Same scene X<α Different scene The above judgment formula was determined through experiments using various characteristic values, and it was confirmed that its judgment accuracy is high. Next, an example will be given and explained. Example 1901 negative films were divided into groups by visual judgment. A: Same scene as the previous frame...361 frames B: Intermediate scene that is judged to be the same scene as the previous frame (similar scene)...540 C: From a different scene from the previous frame: 1000 frames These frames are grouped by calculation. I created three discriminants so that I could separate them. Discriminant that can distinguish it from A frame x ₁ = -0.506 + 4.700 {|ΔRG|+|ΔGB|}+3.21
7｜ΔUL｜+｜ΔRL｝ +4.939｜ΔCP｜+20.515｜ΔDB｜ Discriminant that can distinguish it from B frame x ₂ = −1.765+6.639｜｜ΔRG｜+｜ΔGB}+7 .01
4｜ΔUL｜+｜ΔRL｜｝ +8.300｜ΔCP｜+36.328｜ΔDB｜ Discriminant that can distinguish it from C _top .8
33 {| ΔUL | + | ΔRL | Ta.
The results are shown in Table 1.

[Table] In Table 1, the Mahalanobis distance indicates the degree of separation, and the larger the distance, the better. From Table 1 above, there are 299 identical scene frames A for which visual judgment and calculation judgment match.
There were 294 similar scene frames B and 662 different scene frames C. In the same scene frame A, 163 of the similar scene frames B were erroneously determined as the same scene. However, since this scene can be determined visually as the same scene or as a different scene, no problem occurs even if the scene is determined as the same scene. There were 71 frames in which different scenes were incorrectly determined to be the same scene. Additionally, there were 11 frames that were the same scene but were determined to be different scenes. Therefore, out of 1901 frames, only 88 frames were false judgments. On the other hand, when a conventional discrimination method was used, 278 frames were misjudged. Although the same scene may be determined from the discriminant having the maximum value using the above three discriminants, this would take too long a calculation time. For most of the frames in the same scene, the discriminant x ₁ ≧ discriminant x ₂ ≧ discriminant x ₃ holds, so it is possible to make a determination using the discriminant x ₁ - discriminant x ₂ ≧α . Therefore, the judgment formula is X=1.259−1.939 {|ΔRG|+|ΔGB|
It can be determined using {|ΔUL|+|ΔRL|} −3.361|ΔCP|−15.813|ΔDB|. Zero is normally used as this constant α. When the constant α is set to a negative value, the determination area becomes wider, and even scenes that are not the same scene are determined to be the same scene.When the constant α is set to a negative value, the determination area becomes smaller, and the same scene is incorrectly determined as a different scene. Example In the above judgment formula X, |ΔG|+|ΔYG|+|ΔW|
The frames of the example were classified using a formula with the term added. Here, |ΔG|=|G'-G| is the number of measurement points determined to be green. |ΔYG|=|YG′−YG|, which is the number of measurement points determined to be yellow-green. |ΔW|=|W′−W| is the number of white points. By adding this number of colors, the area of that color is taken into account. For example, |ΔG| is determined based on whether the size of the grass, etc. is the same in the previous frame and the current frame. This adds an additional element for determining the structure of the scene, which improves the accuracy of the determination accordingly. The discriminant X ₂ was defined as follows. X ₂ =K ₁ +K ₂ {｜ΔRG｜+｜ΔGB}+K ₃ ｜｜ΔUL｜
＋｜ΔRL｜｝ +K ₄ ｜ΔCP｜＋K ₅ ｜ΔDB｜＋K ₆ ｜｜ΔG｜+｜ΔY
G|+|ΔW||} The coefficients of the above discriminant X were set and calculated in the same manner as in the example.

[Table] As shown above, when the number of hues is added, there are 302 frames in which the same scene was judged as the same scene, and 61 frames in which different scenes were judged as the same scene. Yes, and there were 7 frames that were judged to be the same scene as different scenes. Therefore, the number of incorrect judgments that became a problem was 68 frames, the number of incorrect judgments decreased, and the number of correct judgments increased to 302 frames. The judgment formula used in this example is X ₂ = 1.33−0.78 {|ΔRG|+|ΔGB|
｜ΔUL｜＋｜ΔRL｜｝ −3.265｜ΔCP｜−15.571｜ΔDB｜−0.0025｜{｜
ΔG | + | ΔYG | + | ΔW |

[Brief explanation of the drawing]

FIG. 1 is a plan view of a negative film, and FIG. 2 is a plan view of a frame showing how the screen is divided.

Claims (1)

[Claims] 1. From the photometric data of the current frame of photographic negative film to be printed, the average transmitted density of the whole area, the average density of the upper part of the screen, the average density of the lower part of the screen, the average density of the right part of the screen, the average density of the left part of the screen In addition to finding characteristic values such as density, minimum density, maximum density, and average contrast of the entire screen, a judgment formula using the above-mentioned characteristic values and the above-mentioned characteristic values of the previous frame consecutive to the current frame is used: X=K ₁ +K ₂ { |ΔRG|+|ΔGB|｝+K ₃ {|ΔUL|+
The value of |ΔRL||+K ₄ |ΔCP|+K ₅ |ΔDB| is determined, and it is determined whether the two frames are the same scene based on the magnitude relationship between the value of this judgment formula and a predetermined constant value. A method for determining identical scene frames. Here, K ₁ to K ₅ : Constant including 0 |ΔRG|=|(LATD′(R)−LATD′(G))−(LATD
(R)−LATD(G)) | |ΔGB|=|(LATD′(G)−LATD′(B))−(LATD(G)
−LATD(B)) | |ΔUL|=|(DL′−DU′)−(DL−DU)| |ΔRL|=|(DRI′−DLF′)−(DRI−DLF)| |ΔCP|=| ((Dmin′+Dmax′)1/2−LATD′(
W)) - ((Dmin + Dmax) 1/2 - LATD (W)) | ΔDB | = Σ (D'i + 1 - D'i) / n - Σ (Di + 1 - Di
)/n (This is the average value of the density difference between adjacent measurement points, and by determining it for the measurement points on the entire screen, it represents the average contrast of the entire screen.) LATD (W): Neutral gray Total area average transmission density LATD(R): Red total area average transmission density LATD(G): Green total area average transmission density LATD(B): Blue total area average transmission density Dmin: Neutral gray minimum density Dmax : Maximum density of neutral gray DU: Average density of the upper part of the screen DL: Average density of the lower part of the screen DRI: Average density of the right part of the screen DLF: Average density of the left part of the screen. Characteristic values without a dash (') represent the characteristic values of the current frame, and characteristic values with a dash (') represent the characteristic values of the previous frame.