Noise reduction system for color television

#This invention relates to color television systems, and is more particularly concerned with signal-processing systems for improving the signal-to-noise ratio of color video signals.

#There is a continuing desire and frequent need in the art of generation and transmission of television to improve the signal-to-noise ratio of the video signal. The problem of excessive noise in the video signal is particularly acute in the currently evolving electronic news gathering (ENG) application of television in which a portable camera is taken into the field, often to locations in which the available light is marginal at best, making it necessary to increase the video gain of the camera to obtain an acceptable picture; this increase in video gain results in an attendant increase in the associated noise displayed in the picture. The problem is further compounded in ENG by the fact that the signal from the camera is usually recorded on a small portable recorder for subsequent playback and transmission by a microwave link to a base station. At the base station, prior to actual broadcast of the signal, the video signal may be subjected to two or three levels of editing, all of these steps introducing additional noise of different forms, so that the ultimate signal available for airing is frequently seriously degraded. Even if the signal delivered by the camera were free of noise, the necessary further processing degrades the signal to a degree that the quality of the portrayal on the home receiver is less than would be desired. A general object of the present invention is to provide a system for reducing noise in a color video signal thereby to improve the signal-to-noise ratio and consequently the quality of the displayed television picture.

#The present invention takes advantage of the fact that television signals are periodic, whereas noise is aperiodic or random. In the NTSC system of television, the lowest repetition frequency is the frame rate, namely, 30 frames per second, with each frame consisting of 525 lines. For the eye to preserve continuity, any two successive frames must be very much alike, and in fact, if no motion is present in the picture, every frame will be a reproduction of the one that immediately precedes it. If a signal waveform is repetitive, signal-to-noise ratio can often be improved by making use of the redundant information inherent in repetition. Systems employing this technique are generally classified as signal averagers, the principle of which is described in an article by Charles R. Trimble entitled, "What Is Signal Averaging?" appearing in the April 1968 issue of the HEWLETT-PACKARD JOURNAL. Although not concerned with noise reduction in color video signals, the paper describes in a general way the principle of noise reduction by signal averaging.

#Other objects, features and advantages of the invention will be evident, and its construction and operation better understood, from the following detailed description taken in conjunction with the accompanying drawings, in which:

#FIG. 1 is a block diagram of a preferred embodiment of a noise-reducing system according to the invention;

#FIG. 2 is a functional block diagram of a circuit for evaluating motion between past and present video frames;

#Although the noise reducing system of the invention is useful for improving the signal-to-noise ratio of the color video signal in any of the currently used color television systems, the invention will be described as applied to the NTSC television, along with appropriate indication of the modifications required to accommodate to other systems, such as PAL and SECAM. Further, although the system may be implemented in either the analog or the digital domain, a digital implementation of the system will be described.

#Referring now to FIG. 1, the video input signal on input line 10, which may be encoded by the pulse code modulation (PCM) technique described in U.S. Pat. No. 3,946,432 utilizing an 8-bit code, is applied via a variable attenuator 12 to one input terminal of an adding or summing circuit 14. The output signal from the summer 14 is applied to a delay device 16 having a delay of 525H - .tau.; H represents one television line interval which means that the delay device introduces a one frame delay since there are 525 lines per frame in the NTSC system. In a practical system the delay is actually slightly less than one frame by the period .tau. for reasons that will become apparent as the description proceeds. The output of delay device 16 is applied through a chroma inverter 18 and a variable attenuator 20 to a second input terminal of the summer 14. Attenuators 12 and 20, shown very schematically in FIG. 1, are ganged and respectively introduce a transmission constant of (1-a) and "a". That is, a fractional portion (1-a) of the amplitude of the incoming or "present" video signal is applied to one input to the summer 14 and a fractional portion "a" of the amplitude of the stored video signal from delay device 16 is applied to the other summer input. It will be evident that if the value of "a" is increased, for example, the proportion of the stored signal applied to the summer increases, and the proportion of the present video signal applied to the summer decreases. Conversely, if "a" is decreased a larger proportion of the present signal and a smaller proportion of the stored signal are applied to the summer.

#Ignoring for the moment the presence of the chroma inverter 18 (that is, assume a direct connection from the output of delay device 16 to attenuator 20), the described configuration is an infinite memory system in which a fractional amplitude portion of the sum of all previous or earlier frames are added to a fractional amplitude portion of the incoming or present video signal, the relative importance of a signal n frames back being determined by the value of "a". For example, if "a" were equal to 1/2 and (1-a) were equal to 1/2, the output signal on the line 22 would consist of 1/2 of the present signal, 1/4 of the next earlier frame, 1/8 of the next frame behind that, 1/16 of the next frame behind that, and so on, with the consequence that the signal seven or eight frames from the present signal has little significance. If there is no motion between successive frames, the video signals representing the successive frames will be signals identical in information content; only the amount of noise in each will differ. When a multiplicity of such identical signals are summed, in the manner just described, the result is a signal identical to any one of the summed signals and of the same magnitude as the attenuated incoming signal by virtue of the fact that the sum of "a" and (1-a) is always unity. However, when random noise present in the video signal, which may vary in amount and distribution from frame to frame, is summed, it tends to be canceled or in any case is not reinforced as is the periodic video signal. It can be demonstrated mathematically that the improvement in the signal-to-noise power ratio achievable, with the described configuration is equal to ##EQU1## Thus, if the value of "a" were 1/2, the value of the fraction would be three indicating a 4.7db signal-to-noise ration improvement. Similarly, if "a" were greater, for example, 3/4, the signal-to-noise improvement would be 8.45db.