JPH02502236A - How to color a black and white scene - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed description of the invention] How to color a black and white scene Technical field The present invention combines manual decisions with information generation and processing. Concerns how to colorize black and white film or videotape films by coloring. be.

Background technology Several devices, including the device disclosed in Applicant's U.S. Pat. No. 4,602.625, Computer-assisted film coloring systems currently exist. The basic principle of the patent is Colors with relatively low information density are created by combining manual specification and computer image processing. generate an information signal, combine it with computer video processing, and then compare it with It is combined with the original black and white information signal with high information density.

Color-only information signals can be transmitted in three ways: low spatial density, low temporal density and and a relatively low information density in any combination with a low color selection density.

inbetweening, cross-dissolving (cross-dissolvlng), techniques like low-pass filtering improve the final result. enhance

The entire process includes the formation of a color-only information signal, the processing of a color-only information signal, and the color The color information signal is combined with the original black and white information signal to generate a full color information signal. It will be explained as including a process.

Computerized colorization of black-and-white film and black-and-white videotape is a recent growth industry. It has become. Several companies now offer commercial coloring services.

However, the products supplied by those companies have been strongly criticized for their low quality. Ru. Therefore, we not only provide coloring, but also provide acceptable high-quality products at reasonable costs. It is desirable to create a device that produces results. This application is filed under applicant's U.S. Patent No. 4. l 30B, some improvements and refinements to 625. improvement is lust Forming, processing, and combining information signals into black-and-white information U.S. Patent No. 4. E108,625 Can be incorporated into the device described or other computer-aided coloring devices .

A particular complaint about commercial colorization is that the color elements of the footage are overly easy and discarded. and the boundaries between colored areas are imprecise, especially when moving quickly. It is inaccurate for the scene in question.

These deficiencies are due to inadequate coloring equipment currently in use. current The current configuration of equipment makes it commercially impossible to produce high quality products.

In order to understand the trade-off between cost and quality, the ``artist'' must It is common knowledge that only the frames are colored and the Combiator J colors all the rest. You must understand that this is a huge exaggeration. Applicant's U.S. Patent 4. Current commercial coloring equipment, including the equipment described in 8O5,625, Human interaction is required to identify and specify the outline of the object to be processed. Man Oral intelligence (AI) and pattern recognition are used to extract known objects in limited situations. It has only had limited success. But they overlap each other to capture a large number of moving, unfamiliar objects that change every few seconds of each camera shot. It is beyond the current state of technology to track and extract . AI research will one day lead to affordable, fully automatic coloring technology; In the foreseeable future, constant human judgment and interaction will be replaced by computers. It is still a necessary component of a coloring device. Therefore, those processes The process is highly assisted by the combination 1, each frame is hand painted. are much more efficient, but those processes are still labor-intensive. It is intensive. The coloring operators are separated and occupy a large proportion of the film frame. You must make a draft of the object to be colored. and labor is a process Since it is the most expensive element, the less objects and colors in the scene, the less the coloring expense. decreases. It is expensive to imitate the visual complexity inherent in "real" color scenes. It takes a lot.

Economy also plays a role in choosing washout pastels. drawing Even with thousands of frames to draw and color, each one takes more time to get the right details. The time to do it is very short. (Furthermore, given that you have infinite time available for coloring However, it still occurs that the boundary of the object differs to some extent from frame to frame. ) And the convenience process of "coloring" the intervening frame is not always it's not correct. These problems become more apparent as the movement in the scene increases. .

When color saturation is low, coloring inaccuracies are less noticeable. a certain transformation In dark frames, the colors are washed out to the point where they are almost invisible. , then neither of them will make a mistake.

Drawing description FIG. 1a shows a typical raw confidence signal frame for a circular region, Figure 1b shows the same frame after low-pass filtering and Figure 1c shows the unprocessed reliability signal line. Figure 1d shows the same cross-section after low-pass filtering, Figure 2a shows the blue A typical image of a red circle on a blue background, Figure 2a, separates the red circle from a blue background. indicates a neutral gray zone, Figure 2b shows adjacent red and blue regions without separating the neutral zone before low-pass filtering. Figure 2c shows the intersection of adjacent red regions separating the neutral zone before low-pass filtering. and indicates the intersection of the blue area, Figure 2d shows adjacent red and blue regions after low-pass filtering without separating the neutral region. Figure 2e shows the intersection of adjacent red regions separating the neutral zone after low-pass filtering. and indicates the intersection of the blue area, Figure 3 shows the output of three polygon drawers (or the output of one or more polygon drawers). shows the output of a videotape recorder (videotape recorder) that plays back recordings. The drawing tool has a circle and a display a limited area that represents the face shown.

First, the facial area is displayed as a pink skin tone.

The second displays the face as blue, and the third displays it as yellow. Those blue and yellow are Used as a modification of the pink signal for shadows and highlights, respectively. three outs The power is sent to a color signal mixer. It is an input end for black and white signal used as control She also has a child. The black and white signal is also mixed with the output of the color mixer to produce a gold signal.

Figure 4a records separate black and white signal components, color tag signal components, and confidence signal components. A digital tape recorder that decodes those components into a color coordinate system such as red, green, and blue. indicates the use of a Figure 4b shows details of the decoder of Figure 4a. Hue generator and color coordinate converter sub section is shown.

Figure 5a shows a face with a uniform flat pink overlay.

Figure 5b shows the same face with a limited area added in a bluish hue to represent shadows. .

Figure 5C is limited by a yellowish hue representing sunlight-lit highlights. Showing the same face with added regions.

Disclosure of invention This application is filed under applicant's U.S. Patent No. 4. [i06.625 and U.S. Application Section 601.091 This is a partial continuation application of No. It is described in U.S. Patent 4.606.625 This invention relates to improvements in coloring devices including coloring devices.

The first set of improvements is based on the boundary contours of objects formed by humans or computers. It is related to reducing the impact of inaccuracy.

As mentioned above, unsaturated colors, i.e. pastel colors, are used for individual images to be colored. Helps hide inaccuracies in boundaries between regions. Also, those defects are further reduced. A set of related color shades is often selected to make the image more appealing. For example, how many In the colored film, most scenes are pink, purple, and blue. It is composed of pastels. Those pastels provide the lowest color contrast selected to do so.

However, the resulting colored film is monotonous and unrealistic.

It may also be intentionally used or recorded as part of U.S. Patent 4.808.625. Alternatively, low-pass filtering, which can be specific to the broadcast process, can be applied to adjacent Color areas can be mixed. Similar pastels blend together without being distracting.

By choosing and using colors that are well saturated and contrasting, A more realistic colored effect is produced, but those colors do not overlap in adjacent video areas. and any inaccuracies in the boundaries between those areas shall be used. , or color mixing is also done more clearly.

Enables the use of more saturated and contrasting colors and A technique to reduce the problem is to change the saturation of video information about the video.

Color saturation is reduced where the probability of error is high, ie near boundaries. certain Where the rate or confidence is high and the color selection is accurate, i.e. away from the boundaries. Colors are more highly saturated in places where

There are at least two different ways to implement the above idea. One is the video This is achieved by forming a "confidence signal" indicating that the color is saturated at each point above. Le. This is done by first marking the boundaries between the individual areas that the computer should color as lines. , for example, by drawing it as black on a white background. those lines Once drawn, the resulting image is low-pass filtered, i.e. “Being blurred.” As a result, a black line surrounded by a blurry gray area You can get an image with The area corresponding to the object to be colored is mostly white. , near the edges of it shades gray. The resulting image is colored When this information is combined with the original black and white information, it is used as control information. control The final golden image is added so that the final image is more saturated when the information is brighter. The color saturation of the image can also be changed.

Another approach is to feed the "confidence" information directly into the color-only information component.

This thickens the borders between individual areas and creates neutral colored lines (generally gray, done by a computer that draws the image as black or white).

When the resulting color-only information frame is low-pass filtered, the neutral wire is buffered. The mixing of colors from two differently colored areas, acting as a color filter. to prevent Furthermore, it shows a “desaturated zone” between the two colored regions. form. This zone varies smoothly with minimal saturation at the center of the line.

An additional result of either of those realizations is that when color-only information is combined with black-and-white information, In contrast, the desaturated zones around the edges of objects are reduced by the color overlay. It is a departure from the edges of black and white objects whose obscurity has been reduced. Perceive edges High-density visual information in a scene, such as that used by the brain during sleep, becomes a black-and-white information component. Supplied by Therefore, the black and white edges are less obscured by color. When separated, the colored film appears soft or fuzzy compared to the original black and white film. Helps reduce the awareness that there is.

Draw the line used to generate the confidence signal or the neutral boundary of the second realization Time can vary the quality of those lines based on several criteria. change Possible qualities include line width, opacity, darkness, or tint. Also, those changes the amount or quality of low-pass (or other) filtering applied to the picture containing lines of can be done.

The criteria for changing the line or filtering quality are the characteristics of the individual areas to be colored, or It can be based on the characteristics of two adjacent areas that form a boundary on a line. For example, the dimensions of an object - Very small objects can be bounded with thinner lines or filtered less. It is necessary to

Object movement - Edges of fast moving objects are less distinct and more difficult to specify. Since such objects require thicker lines and more or less filtering It requires at least one of the following. Objects at rest require thin boundaries . The reason is that if you look at thick boundaries for a long time, you will notice them.

Object prominence – Objects that stand out or are important to a scene require special treatment. may be required. Or objects can be made to stand out through special processing.

Luminance contrast between objects - the brightness between two objects separated by a line The line and filtering quality can be changed according to the difference.

Chrominance contrast between objects - between two objects separated by a line The quality of the lines and filtering can be changed according to the intended difference in color tone.

Confidence in the specification of object boundaries - for example, if the boundaries of an object in a frame are determined by human manipulation , and the boundaries of objects in different frames are interpolated by the computer. If the angles of the two types of boundaries are not equal, the angles of the two types of boundaries may not be equal. Therefore less accurate boundaries may be represented by thicker lines or differently filtered.

The above examples are intended to be illustrative rather than limiting in nature. It is. line parameters and filtering parameters, or other video processing parameters. Other criteria can be considered when adjusting the data and are within the scope of the invention. Ru. In general, the line and filter parameters should be adjusted to give the best perceived results. data is adjusted.

For example, if those techniques are applied too sparingly, Problems with the intention to do so will remain. that those techniques were strictly applied; These techniques can then produce their own visual abnormalities. (for example, Lines that are too large and filtered too little will create a perceptible uncolored border around the object. I sometimes cheat. ) those parameters are adjusted to achieve optimal parallelism. Ru.

The next set of techniques is related to more efficiently contouring the boundaries of objects. It is.

The practical problems that exist in forming the boundary contours of objects with traditional computer-aided coloring techniques The subject is The human labor part makes the process more expensive and "draws" enough boundaries to look like the real thing. That is unrealistic.

Humans and computers are not mechanically accurate, but follow the same ``edge specification rules.'' cannot be applied consistently across systems.

On the other hand, fully automated computer equipment is very cheap and works very accurately. However, AI computers can also identify arbitrary objects or the boundaries of those objects. It is not possible. The following describes ways to improve the effectiveness of human and computer efforts. in a description of a device that combines those two ways to achieve a combination. This means redefining the division and overlap of assigned tasks. operator wear You will be freed from the burden of repetitive work. Identification beyond the capabilities of computers The role of the operator in making decisions will increase.

The interface between operators and computers has become more flexible and complex. This allows humans to more fully convey their judgments to machines. So Computers can then make these decisions far more efficiently than humans can. Work efficiently.

This device consists of a collection of algorithms for edge and contour extraction and a set of line editing Contains two main subsystems with functions: Convenience store is such software. It has the ability to apply tools efficiently and quickly. However, now Computers have developed tools to best apply these "tools" to black and white video frames. It does not have the ability to determine the cutting edge axis and location.

(Therefore, the computer can store the color information required to color a black and white frame.) cannot form object boundaries that include frames. Operators can easily issue complex commands Other functions in the total interaction structure are divided into two main subsystems: integrated into the stem. It is adapted to be used for a series of frames.

Combiator video is well known for extracting contours and edges from digitized video Processing technology exists. They are based on differences in brightness, texture, sharpness, etc. including extraction based on connectivity, or other criteria. those The technical details of are not within the scope of this invention.

In this paper, we use a code to quickly and cheaply form potential boundaries between objects. Edge and contour extraction algorithms are applied by the computer. large numbers These lines are "false boundaries." For example, superimpose on one or more objects You can draw an outline of a shadow as a clear object. Similarly, the brightness of adjacent objects , two objects cannot be completely distinguished if their textures, etc. are very similar. There is.

In this respect, the scene in which various edge and contour extraction algorithms are superimposed is Displayed for the operator. Edit all potential boundaries gracefully and efficiently A set of software functions is provided to the operator to enable the operator to do so. Those functions are integrated into the interaction device. You can eliminate some lines and use other lines to connect objects. Allows connections to continuous boundaries, adjusts lines, and eliminates gaps or missing boundaries. Features and other line editors that allow you to "hand-draw" the parts that fill the field Includes ability.

For the first frame of a given scene, the operator and computer Interacts through a set of software functions to control which video processing algorithms how to apply it to the video area and how to set the parameters of each algorithm. The operator instructs the computer as to how to make the adjustment. extract as result Displays the line placed on top of the line and allows the operator to set various parameters in the function menu. This can be done by changing the result for the menu. This is a video processing About the rules for applying algorithms to several similar frames in a scene state

The learning involved in operating the above device is comparable to that of “traditional computer-aided” coloring devices. Requires more training, experience, and effort on the part of the operator compared to do. However, compared to computer-assisted contour handwriting, the efficiency, The long-term gains in accuracy and consistency are more than worth the extra effort. It is something.

Operate how to best apply various video algorithms to a scene. The controller instructs the computer to create a If you edit the potential border that was created, the second frame and subsequent frames will look better. processed in an automated manner. The object as a whole is very small for each frame. There is a kana movement. Therefore, the extracted lines of the second frame are The approximate location and shape of each line will be similar, although slight changes may be made to the system. However, retains the edits made to the first frame by the operator and The computer includes software for applying the same functionality to the next frame. . For example, "Exclude the 4 long lines in the top left corner and make the 3 short lines on the right side into a continuous border" The operator edits the border frame for the first frame in order to then this same sequence of operations is retained and applied to the next frame.

The exact position and shape of each line can be changed slightly, but it is It does not exceed the capabilities of current AI to take into account.

A combination of precise line shaping and editing functions that apply to each subsequent frame The operator has the opportunity to investigate the attempt. If the computer makes a mistake, the operation The editor has the ability to re-enter line editing mode and make corrections. Those modifications will be retained. to update the editing feature set so that it is retained and automatically added to the next frame. It is used for (Similarly, at any point in this sequence, the line extraction algorithm Rhythm is specified and its rule set is updated as it is added by the computer. You can enter the mode where ) As mentioned above, the additional features of the software are Information about the movement of the object is retained and the computer uses it during line extraction and line editing. Used to update the rule set to be applied by for example," Image processing algorithm that extracts lines by comparing textures in the first frame Assume that the operator instructs the computer to apply the . The lines are formed by an extraction algorithm and then compiled into object boundary specifications.

For subsequent frames, the boundaries of the object formed by this process will be If it moves at the same time, the "line extraction by texture comparison" algorithm is applied. The area that is moved is moved in the same way.

Similarly, for the line editing stage, you can connect several lines and connect all other lines that are stepped in the boundary. If the boundary of an object is formed by deleting all lines, then the edited As the boundary of the object is moved from frame to frame, the associated line is moved from frame to frame. The computer automatically checks the updated position for editing in the frame. commanded.

What has been described above includes automatic object tracking.

Another way is to apply a combinator interpolation intervention. Object position, argo The operator does not adjoin the area to which the rhythm is applied or the associated line to be edited. Specify in two frames. The computer calculates the intervening position and Apply the rule to the existing frame.

As AI software becomes increasingly sophisticated and capable, the above adaptations Functions become increasingly automatic and accurate. Eventually, the operator will Color and outline only the frame (or do it on every frame) ), it is possible to develop a coloring device that truly works. The above process is like that It forms the basis of a fully automatic device, which is within the intended scope of the invention. ~ Relevant features of sophisticated equipment are often techniques applied to the image features encountered. be. For example, the problem mentioned for a particular colored film is ``01'' blue. Frank Sinatra (01'B1ueEye) RA's well-known nickname) is depicted as ``01' brown eyes.'' this is , it is possible to draw the contour extraction algorithm separately for Sinatra's blue eyes as individual objects. due to the fact that it is not practical.

The new tint applied to the rest of his face is applied to his eyes as well, making them darker. Forms deep pink or "brown" eyes. The outline of hibernation against the face in each scene would place an impractical burden on current coloring equipment. Also, the current Automatic identification of eyes in a scene is impractical with current AI technology. However, people and Devices based on machine cooperation are practical and will be described below.

The operator has some additional functions in his menu, e.g. "Eye Identification". Ru. The operator can then control the location of the keys, such as the left corner, right corner, and pupil. point it out to the computer. When the computer is told the appropriate image area, it is well defined. Based on the established feature extraction rules, the computer determines the various parts of the eye that should be colored individually. Identify and specify minutes. The feature extraction rules are ), the relative darkness of (Caucasian) skin tone, and the relative brightness of the eyes. It can be based on the round shape and the long, vague shape of the eyebrows. Then, the membrane, Individual eye characteristics such as pupils, eyebrows, and eyelashes are individually specified by the computer. and the appropriate color and other areas, such as the eyelids, are treated in the same way as the facial skin.

Similar features enable collaborative (human-computer) identification and Allow processing to be performed as part of the "mouth identification" function. hands, shoes, tie, hat , hair, etc. Other common object or image features can also be used. specific fill Specialized functions for movies, such as guns, trees, birds, etc. in western movies. It can include things like mountains etc. ``Object tigers'' similar to those mentioned above. At least one of the "locking" and "detection by intervention" mechanisms can also be can find its own such object to interpose the frame It can also be incorporated into "feature identification" software. The above example is limiting It is meant as an example, not as a property.

The above features of the sophisticated device reduce the inaccuracy of the object boundary contour and reduce the object boundary circle. This relates to forming the shell more efficiently. The next few refinements are Washed out, too easy, not advanced, not complex, or too scary This relates to improving the appearance of the coloring mentioned above.

As mentioned in the above criticism, the colored films currently produced are not interesting. It's not very sophisticated.

In its most limited use in the color selection area, applicant's US Pat. No. 4. 110B, 625, produces similarly uninteresting results, but it occurs much more quickly and much more efficiently. '625 as taught by In addition, due to the limited color selection density j of the color signal, the amount of time, money, processing, and and the use of labor-saving techniques, or to display the final color-only signal. This allows the use of high speed polygon drawing devices.

However, by keeping the color-only signal in this "low color selection density" state, the entire coloring process is It is possible to obtain these benefits across all stages. Moreover, this process In particular, in the final stage of the process, the color-only information signal, the original black and white high-density information present in the black-and-white video to which it is combined. can be modulated or changed based on information. The resulting modification is the change By adding precision and precision, the information signal of only color is strengthened, making it more interesting and realistic. It produces a great effect.

In its most specific embodiment, r color selection densities J of the color-only information signal are surrounded by Limit to one color for each region. U.S. Patent 4,608,625 flat The result is a simple colored film. However, each color is uniform. Such a representation as an enclosed region can be interpreted as input to a fast polygon drawing device. imaginatively suitable. Developed for flight simulators and CAD/CAM equipment These devices can draw colored polygons extremely quickly. that Some of these devices use colored areas contained in frames of only one color. in real time and near real time (i.e. 24 or 30 times per second) can.

The disadvantages inherent in using a polygon drawing device as described above can be overcome by the following method. can. We will limit the discussion to one enclosed area representing a human face. above , the rendering device displays one uniform pink enclosed area. However, this It is advantageous to change the color of the face over the area. For example, if the sun is shining on your face. When exposed to bright light, the brightly lit areas of the face are colored a slightly yellowish peach color. It can be shown by

Similarly, heavily shaded areas are colored a slightly bluish pink. child This effect can be obtained by operating three drawing devices in parallel. High density black and white emotion The information signal is also input to the video mixing device as a control signal.

When the control signal is the brightest, more and more yellow signals are added to the pink signal. . When the control signal is the darkest, more and more yellow signals are added to the pink signal. The full tonal range of the skin is produced, from midtones to yellowish highlights. Ru. This decoded signal is combined with a high-density black and white information signal to create a more realistic golden image. Form. The above mixing equipment is a well-understood analog video, post-production This can be realized using post-production technology.

In another embodiment, a single rendering device can be used to record three color signals to videotape. One color signal can be displayed at a time in order to This is practical. That's a picture This is because the imaging device operates very quickly. 3 @ videotets instead of 3 drawing devices Play audio tapes synchronously.

Four or more drawing devices (to generate four or more uniform color signals for mixing) or videotape). Furthermore, the visuals of only those colors are The method of mixing video signals should be simple and straight forward as described above. There isn't. Using many types of complex mixing techniques, the final product is made with great variation and great sophistication. It is possible to form a realistic image.

However, if the image is complex enough to require more than four rendering devices (or videotapes) If it were desirable to form a color mixing technique, this technique would be difficult to use. Alternative embodiments are preferred. This alternative embodiment requires one drawing device, but more Requires specialized and sophisticated mixing equipment.

In this example, each enclosed area is painted with a unique uniform color. This output is input to the mixing device along with the high density black and white information signal. For each unique and uniform color Therefore, the mixing device replaces the color depending on the luminance value range of the black and white information signal at each point. use a different color. The number of brightness value ranges can be 3.5 or 3.5 depending on the requirements and precision of the mixing device. can do more than that. The range difference changes smoothly and continuously for more modulations. You can get it.

Preferred embodiment Introduction of new equipment, namely Sony's digital tape recorder (DTR) As a result, it has become convenient to combine several of these precise functions into one device. It was. DTR devices record three independent digital information signals. each digital The information signal is a monochromatic signal.

Y (brightness), Y-R (red minus brightness), and B-Y (blue minus brightness) ). Initially the signal is recorded at twice the density of the other two. (each signal is 1 Record independently with one device or reciprocate back and forth between two DTRs. By doing so, it is possible to insert each signal independently. DTR is a digital record Since it is a digital camera, there is almost no signal deterioration when recording a large number of generations. stomach. ) Therefore, to realize the apparatus described in U.S. Pat. No. 4,806.625, DTR is ideal. The original black and white information signal is stored with higher information density and stored in the channel. The other two channels are used as follows.

To record the output of a frame buffer or the output of one fast polygon drawing device One channel is used for each. In this example, each individual region has one Unique and uniformly displayed as any "color". The signal recorded by DTR is Since it is a monochromatic signal, only shades of gray are recorded. Therefore, all shown must be shaded with gray. Various enclosures to be colored separately within the picture Their gray "color" will be used later to distinguish between the areas that have been added. This signal is , once recorded, is called a "color tag signal".

The other channels are used to store reliability signals as explained at the beginning.

If these three types of signals are played back using a normal playback circuit of a recorder, the result will be I don't understand. However, the output of the three types of signals is instead a specialized sent to the device numbered. The output of its backward motion is very delicate and realistic golden signal. .

The decoder itself is a very precise digital device, with its own microprocessor. It has a processor and a memory circuit, and operates as follows. Three digital inputs to the decoder The power is the original black and white information signal with high information density reproduced by DTR and the original black and white information signal with low density. They are a "color tag signal" and a "reliability" signal.

The hue of each point on the screen is determined from the combination of the color tag signal and the original black and white signal. Extracted. The information stored in the memory of the decoder is for each color with different black and white signals. Specifies how the tag is modulated. Once a shade is selected, a “Confidence” confidence is displayed. The color saturation is changed by the corresponding part of the code. Finally, each Luminance values are extracted for the points. The hue, saturation and brightness values are determined for each point. Once determined, the final stage is to set the appropriate configuration for whatever particular recorder is being used. The goal is to convert those parameters accordingly. They are R, G, BSY, R -Y, B-Y or other suitable color specification varata.

A number of advantages are derived from this embodiment that combine to produce a precise golden signal. and/or independently store various information signals. The goal is to ensure that Furthermore, two color component signals (color tag signal and confidence signal) can be formed and stored separately with very low information density and in the most convenient manner.

In order to combine three types of signals for each frame, the main computer has storage space and processing space. There is no need to provide administrative time. This combination can later be combined with an inexpensive and real-time decoding step. It will be done as follows.

Two low-density color component signals can be formed and stored with very low information density, and high-density It is necessary to combine those component signals until the last stage to form a golden signal There isn't. Therefore, high-density full-color signals or high-density intermediate signals can be stored in computer memory. There is no need to store the signal on film or videotape at the final stage. It only needs to be recorded.

Next, U.S. Patent No. 4,606. Another modification of a lower degree than the device taught in H5. It also allows for film coloring that is superior to current commercial products. . Although less technical than the preferred embodiment, this technique has its own advantages and ' 825, which is applicable to the preferred embodiment described above and to current commercial equipment.

The goal of this technology is to create a more refined color signal, but the conventional technology Differently, it can be done by increasing the information content of the color-only signal at the beginning of the process. I do. However, the amount of additional information is limited by the amount of additional information that is formed in the final stages of the preferred embodiment. There is little color information. Therefore, this additional information can be used in processes (e.g. Additional burdens are acceptable throughout all stages of processing (image processing, processing, storage, etc.). Additional information is formed completely and automatically by computers, and the overhead of human labor does not increase. stomach.

Basically, this technology is specified by human-machine interaction or other processes. Includes automatic identification and specification of subregions within enclosed regions. their subregion boundaries The fields are contoured from density cuts or other criteria.

For example, going back to our example of a face illuminated by sunlight, the values of the face range from all (0) to all (0). Assume that it changes to part (1). It also has a uniform pink color to show the effect of sunlight. Suppose you want to create yellowish highlights and bluish shadows instead of a face of Ru. In this technique, separate boundaries of the shadow sub-regions are formed as follows. combination Twenty-one nights are set at a certain threshold, for example 80. All of them within the boundaries of the lower facial region is instructed to consider every pixel. The computer is further forming a sub-region boundary around the connected group of these pixels. commanded. Computers tend to ignore pixels that are too small to be important. It is also ordered. The result is an island of shadow that is contoured within the boundaries of the face. those The islands are now considered to be separate enclosed areas, colored bluish-pink. It will be done. Similarly, highlight islands (e.g. pixels higher than value 210) are yellow Treated as a separate enclosed area colored a pinkish tint.

Alternatively, their subregion boundaries may be pixels of just one value or a narrow range of values. It can be formed by extracting the contour defined by the cells.

Sub-region boundaries can also be formed based on other criteria. For example, subtlety and Subregion textures or parameters that are colored separately to add subtlety. data can be disassembled. For example, one enclosed area by a human interface Let's consider the wall as The wall is covered with parameterized wallpaper , we want to find various elements that repeat (e.g. strips, flowers, etc.) or backgrounds. , and forming a separate colored subregion border around each element. You can give commands to your computer. fabric, stones or bricks on the wall, flowers from the garden, or any number of objects The body can be drawn in any other textured area where it would be impractical to draw by hand. Similar techniques can be applied to area strips or grids.

Such examples are meant to be illustrative rather than limiting; Variations on this technique are intended to be within the scope of this invention.

Next, commercial coloring equipment is used to obtain clean, restored black and white prints before coloring. Unusual attention was paid to Cleaning the negative and making new prints Although it is not a complicated task, scratches and hard-to-remove stains may occur on new products. White (or black) dots or lines may also appear on printed prints. US Patent 4, [101! , 625 form a color-only information signal. Sometimes the original black and white footage is used as a reference, but it remains unchanged until the final combination stage. It is.

In some embodiments, the signal hue of only the color varies, but the luminance value is uniform and moderate. It is used simply as an overlay signal. For those examples or other similarly functioning coloring devices, apply the techniques described below to their implementation. It can be incorporated into the example.

Although a black and white signal can be used to form a color signal, the operator may Note scratches, holes, and other white defects in the statue. Then the operator does this Select the repair function from the menu and then tell the computer the defect.

Then, the computer will have the appropriate hue in the color-only information signal, but Create a patch with a darker value that corresponds to the area surrounding the scratch. In addition, In contrast, the edge between the enclosing areas in the color-only signal is made softer. to hide the fact that there is a ``next'' part.

Similarly, for dust defects or dark defects on black and white frames, the operator , select different repair functions to form a brighter color sequence in a color-only signal.

However, darker patches can cover white holes better, while lighter patches are less effective. dark defects, but not completely eliminate dark defects.

Therefore, among the things clarified from the explanation, the above-mentioned various points can be efficiently obtained. and in carrying out the above method and making certain changes in the structure. I can. Therefore, as contained in the above description or shown in the accompanying drawings, All matters contained herein are illustrative and should not be construed in a limiting sense. .

Having described the invention above, it is assumed that it is new and that it is desired to be protected by a letter of patent. The following items will be charged:

tnla diagram mlb diagram jf) I CrXJ Figure 1d Color signal contour database Figure 4a Figure 4b

Claims (24)

[Claims] 1. a. within a black and white video frame to form at least one designated area specifying at least one region of b. generating a border around at least one of the designated areas; c. processing the boundary line to generate a graded information signal; d. Adding the graded information signal to the color information signal produces graded sensuality. The process of forming a signal A method for generating an information signal comprising: 2. Claim 1: The process specified in step c is based on the characteristics of the specified area. Method described. 3. said graded information signal to form a graded color information signal; 2. The method of claim 1, wherein the step of adding the code is performed as part of the decoding procedure. 4. a. identifying at least one region in the black and white information signal; b. separating the black and white information signal into at least two designated signal portions; , c. Forming a color-only signal portion for at least two of the specified signal portions d. Separate the signal parts of at least two colors by a neutral zone. The process of e. The color-only signal portion is combined with the monochrome information signal to form a full-color information signal. A method for coloring a monochrome information signal. 5. Claim: A step of processing the color-only information signal is added between steps d and e. The method described in 4. 6. The quality of the neutral zone is determined according to some characteristic of at least one of the signal portions. 5. The method according to claim 4, wherein: 7. varying the quality of the processing according to some characteristic of at least one of the signal portions; 5. The method according to claim 4. 8. a. a process of interactively adjusting parameters of an image processing algorithm; b. forming a large number of video lines using the video processing algorithm; c. a process of editing the video line to form boundary information; A method for generating boundary information comprising: 9. Applying the adjusted video processing algorithm to a number of videos in a video sequence. 9. The method of claim 8, comprising the additional step of adding. 10. Editing functions on video lines formed from multiple frames in a video sequence 9. The method of claim 8, comprising the additional step of adding . 11. adaptively adjusting the video processing algorithm based on the movement of certain video elements; 9. The method of claim 8, comprising an additional step. 12. an additional process for adaptively adjusting said editing functions based on movement of certain video elements; 9. The method according to claim 8. 13. a. a process of identifying objects in a black and white video frame; b. inputting the location of at least features of the identified object into a computer; c. Image processing algorithms are used to generate boundary information about the image object. the process of instructing a computer to add A method for generating boundary information for a video object comprising: 14. 14. The method according to claim 13, wherein the imaged object is an eye. 15. a. black and white video frames to form at least one designated area; specifying at least one area within the b. forming at least two color information signals for the designated area; , c. at least one based on at least one video feature of the designated area. modulating the color information signal of the color information signal to generate at least one modulated color information signal. process and d. combining the color information signals to generate a composite color information signal; A method for generating a composite color information signal comprising: 16. 16. The method of claim 15, wherein the video feature is brightness information. 17. 16. The method of claim 15, wherein the video feature is texture information. 18. Separate information including a black and white information component, a color tag information component, and a reliability information component A series of black and white video frames employing a digital tape recorder to store the components How to color your own. 19. a. black and white video frames to form at least one designated area; specifying at least one area within the b. to generate an information signal delineated by a polygon for the specified area; a process of using a polygon drawing device; c. The information signal described by the polygon and the black and white image are combined to generate a color information signal. a process of combining and processing images; d. combining the color information signal and the black and white image to generate a full color information signal; 1. A method for generating a full-color image, comprising: a processing step; 20. a. black and white video frames to form at least one designated area; specifying at least one area within the b. generating a border around at least one of the designated areas; c. processing the boundary line to generate a graded information signal; d. generating an information signal depicted by a polygon with respect to the specified area; , c. The information signal described by the polygon and the black and white image are combined to generate a color information signal. a process of combining and processing images; f. said graded information signal to generate a graded color information signal; adding a color information signal to the color information signal; g. Combine the color information signal and the front foot black and white image to generate a full color information signal 1. A method for generating a full-color image, comprising: a processing step; 21.8. black and white video frames to form at least one designated area; specifying at least one area within the b. to generate an information signal delineated by a polygon for the specified area; a process of using a polygon drawing device; c. The information signal described by the polygon and the black and white image are combined to generate a color information signal. The process of combining and processing images A method of generating full-color images with 22. a. black and white video frames to form at least one designated area; specifying at least one area within the b. specifying at least two sub-regions for the designated region; c. specifying a color selection for each sub-region. 23. 23. The sub-area is specified based on the luminance characteristics of the specified area. the method of. 24. a. identifying at least one region in the black and white information signal; b. separating the black and white information signal into at least two designated signal portions; , c. one color-only signal portion for at least two of the specified signal portions; a process of forming; d. determining at least one video defect in the black and white information signal; The process of e. In order to correct the video defects identified above, a change in density is created in the white-only signal. f. the color only signal and the white to generate a full color information signal and combining black images.