FR2534435A1 - Single-sensor colour television camera. - Google Patents

Abstract

The single-sensor colour television camera comprises a solid-state imager 20 with line transfer, delivering on a first and on a second output two signal trains corresponding respectively to the odd and even points of the rows of the matrix of photosites of the imager, this imager being associated with a chromatic filter 10 for which all the elementary filter lines of the same frame of the matrix are similar, alternately green and blue for one frame, and red and green for the other frame, the elementary green filters being staggered. The formation of a TV image is obtained by reading the entire imager for the duration of one frame of the TV image, two successive lines of the imager being read for the duration of one TV line and combined in the associated processing circuits 30, to obtain the three components of the colour TV imager.

Description

Monosensor color television camera
The invention relates to a monosensor color television camera. The production of such a camera requires the association of a solid-state imager, a mosaic of colored filters and a processing channel for video signals from the imager. There are currently many solid-state imager structures that can be used to make a color monosensor camera.

The invention relates more particularly to a monosensor color camera built around a line transfer imager, for example an imager of the THX 1139 type manufactured by the company THOMSON
CSF. This type of imager is interesting because it has good sensitivity in blue, the possibility of anti-blooming control, a less complex structure than an inter-line transfer imager, and low temporal noise.

 The problem which arises is to build an optimized camera with such an imager by associating with it the chromatic sampling structure, that is to say the mosaic of color filters, the most satisfactory, and by also associating with it a channel of treatment as simple and efficient as possible.

 There are many structures of colored mosaics which appear fairly close to one another and provide suitable results. The monosensor color camera according to the invention comprises a colored mosaic associated with a solid state imager as defined above operating according to a mode hitherto not provided for by the manufacturer, associated with a particularly simple processing path.

According to the invention, a monosensor color television camera, is characterized in that it comprises in combination:
a solid state line transfer imager having two outputs delivering signals associated respectively with the odd points and with the even points of the successive lines of the photosite matrix of the imager,
- a chromatic filter composed of associated elementary filters
photosites such as filters of the same rank in successive rows
ves of a matrix frame are similar, alternating filters
green and blue being performed on the lines of a frame and alternating
of red and green filters being carried out on the lines of the other frame, the
green filters staggered over the entire matrix,
- and circuits for processing the output signals from the imager
to form the three components of a color TV image, the formation
of this image being obtained by controlling the reading of all the
lines of the imager successively for the duration of a frame of
the image, two successive lines being read for the duration of a TV line
and combined in the processing circuits to form the components
of a line of the TV picture.

The invention will be better understood and other characteristics appear.
will reiterate using the following description with reference to the figures
attached:
- Figure 1 is a block diagram of the following camera
the invention;
- Figure 2 is a diagram of the imager used;
- Figure 3 is the diagram of a known colored mosaic, known as -type BAYER;
- Figure 4 is the diagram of the mosalque used in the camera according to the invention;
- Figure 5 is an explanatory diagram for the intertwining of lines;
- Figure 6 is a time diagram of the control signals applied to the imager;
- Figure 7 is the diagram of a first embodiment of the camera processing circuits according to the invention;
- Figure 8 is the diagram of a second embodiment of the camera;
- Figure 9 is an explanatory diagram of the operation of the
camera.

In Figure 1 which shows a block diagram of the camera
monosensor according to the invention, a color filter 10 is placed on the front face of a line transfer imager 20 of the THX 1139 type, this imager having two outputs delivering signals Si and S corresponding ip respectively to the odd and even points of the imager matrix, these signals being applied to processing circuits 30 delivering the camera color signals, R, G, B, or Y, DR, and DB.

 FIG. 2 represents a block diagram of the line transfer imager used in the camera according to the invention. This imager comprises a photosensitive zone 1 consisting of a matrix of m lines each comprising p photosites, two logical shift registers for selecting the lines, 2 and 3, one associated with the odd lines of the matrix of the imager, l other to the even lines of this matrix. This matrix comprises, in the described embodiment, 576 lines of 462 photosites.

Each of the selection registers comprises m2 selection outputs, each of these outputs being connected to all the photosite selection inputs of the same row of the matrix. Two successive lines of the matrix are controlled respectively by the register 2 and by the register 3, to produce an alternation of lines forming a frame called odd and a frame called even, interleaved. The photosites of the same column of the matrix are linked by. a conductive column to a first or to a second row memory, 4 and 5, depending on whether these columns connect photosites of odd or even rows in the rows.

The memories of. lines 4 and 5 are respectively connected to two shift registers with transfer of output charges 6 and 7, the output signals of these output registers being respectively the signals S. and Sp correspond
1 spawning at the odd and even points of the photosite matrix.

 This imager works as follows. The lines of the two frames of the photosite matrix are selected, according to the frame parity, either by the selection shift register 2 of the even frame on command by the control signal C52, or by the line selection selection register the even frame on command by the command signal CS3. After selecting a line, the charges stored on the photosites are transmitted to two line memories 4 and 5, the charges of the photosites of odd ranks being discharged in memory 4 and those of the photosites of even ranks in memory 5. These charges are then transferred into the two charge transfer shift registers 6, 7 and then these two registers are read at a TV rate to deliver the content of a video line.

The chromatic sampling structures conventionally associated with an imager of this type are mainly BAYER type structures. Figure 3 shows such a structure. In this type of structure, the colored filters placed on two successive lines of the photosite matrix are identical, the even and odd frames of the lines of the matrix giving the same type of video signals. Given the fact that the luminance is mainly present in the green components, this component is preferred and the alternation of the filters on two successive lines of the same frame is as follows:
... VRVRVR ...

 ... BVBVBV ...

 As can be seen, the green samples are staggered on these two successive lines of the same frame. This type of structure is generally used in the following way: each line of photosites corresponds to a line of the image, the even and odd frames of photosites corresponding to the even and odd frames of the image. To obtain the three colors simultaneously, it is necessary to use a line delay line because there are only two of the three components on the current line, V and R or V and B. green samples on two consecutive lines of a frame being staggered, the improvement of the horizontal resolution is achieved by multiplexing these samples. All the processing carried out therefore results from combinations between two consecutive lines of the same frame.

But these two lines are not often correlated with each other, especially in the presence of oblique transitions, due to their spatial distance because they are separated by a line from the other frame. In addition, multiplexing, which improves the horizontal resolution, simultaneously causes a degradation of the vertical resolution.

FIG. 4 represents the sampling structure used in the monosensor color camera according to the invention. In this structure, all the lines of the same frame of the photosite matrix are covered with the same colored filters, ie BVBV ... for the first and
VRVRVR ... for the second. Thus, the green samples are preferred, as in the BAYER structure, but two successive lines of the matrix include the three colors. Digital processing has been chosen, and, for cost reasons, it is advantageous to use an analog-digital converter at the output of the imager having as low a resolution as possible, the images obtained however not being too degraded. by quantization noise. This is why it is necessary that each of the three components R, G, B, be coded with the maximum resolution, which would not be the case by using chromatic filters with complementary colors which result from the addition of two or three primary colors.

 This chromatic sampling structure is such that the distribution of green samples is optimal on both axes, horizontal and vertical.

 This color filter being associated with the line transfer imager, it is then possible to improve the resolution to work simultaneously, to obtain a line of the image, no longer on two consecutive lines of the same frame. but on two consecutive lines of the matrix of photosites each belonging to a frame. For this, it is necessary that these two successive lines be read for the duration of a line of the image, that is to say 64 micro-seconds, these two lines belonging respectively to the even frame and to the frame odd from the photosite matrix. Whenever a line is selected, the 231 odd points of this line are stored in memory 4, and the 231 even points of this line are stored in memory 5. line of the image of two consecutive lines of the matrix of photosites, it is necessary that the output register delivers the 231 points of the first line then the 231 points of the following line during the duration of a line, minus the times transfer of the first and second line in the memories.

Given the signals available in the camera, it is particularly interesting to have a clock frequency, for shift registers with output charge transfer, equal to twice the PAL subcarrier, i.e. - say at 8.86 MHz; at this frequency the sampling of the 231 even or odd points of each line can be carried out in 26 microseconds, which leaves 6 microseconds for the transfer of photosensitive element-element to CCD.

 The timing diagram of the control signals of the various elements of the imager for this new operating mode associated with this new sampling structure is shown in FIG. 5.

 It has been indicated above that, in order to obtain a good quality image, the processing operations were carried out from the samples corresponding to two lines that are spatially correlated, that is to say two successive lines of the matrix of photosensitive elements. It is therefore necessary to obtain each frame, whether it is even or odd, to read the whole of the imager. A line Ln of the odd frame of the image is produced from a line ln of the odd frame of the imager and a homologous line In ′ of the even frame of the imager. The line Ln 'of the even frame of the image is produced from the line-ln' of the even frame and the line ln + l of the odd frame of the imager. Figure 6 is an explanatory diagram which shows how to group the lines to obtain respectively the even and odd fields of the image from the lines of the photosensitive matrix. This operating mode is controlled as indicated above by the signals shown in FIG. 5.

 Between the beginnings of two line blanking intervals, two successive lines from the imager are read. For this, the selection register of the first line, namely the odd line, is controlled by the control signal Cs2, then after a line half-period, the selection register 3 is controlled by the signal C53 for the reading of the next row of the matrix. The total duration of each reading does not exceed 6 microseconds, and in the line half period there remain 26 microseconds to sample the even and odd points of the first line then of the second line respectively in the output shift registers 6 and 7.In this figure have also been shown the controlled transitions on the electrodes of the various elements of the photosensitive matrix, VST being the signal applied to the electrodes corresponding to the memory elements, Vp being the control signal applied to the electrodes which allow the drainage of the charges to the memory elements, and c being the signal applied to the electrodes for controlling the transfer of the charges stored under the electrodes from the memory elements to the output register. Finally, the HT and HD control signals are applied to the charge transfer registers for the transfer and offset of the charges in these registers.

 The information from the imager is transmitted to the processing circuits 30 in FIG. I. A first embodiment of these processing circuits is represented in FIG. 7, a second embodiment being represented in FIG. 8. The distinction which exists between these two embodiments lies in the proportion taken by the analog processing before the analog to digital conversion.

 In FIG. 7, the outputs Sp and Si of the imager 20 are connected to the inputs of two pre-amplifiers 31 and 32 respectively, the outputs of which are connected to the inputs of a multiplexer 33. The output of the multiplexer 33 is connected to the first input of a differential amplifier 34 whose output is connected to the input of an analog-digital converter 35. The multiple output (the analog samples of the signal being coded on eight bits for example) of the analog-digital converter is connected through a demultiplexer 36 at the inputs of three memories 37, 38 and 39 with capacities equal to at least two image lines. These memories can have a capacity of 512 x 16 bits. The multiplexer 36 also has an output connected to the input of a digital level alignment and noise correction circuit with fixed structure 40, the second input of this circuit. being connected to the output of the memory 39 in which the digital values corresponding to the characteristic samples of the noise with fixed structure are stored.

The output of this digital level alignment and correction circuit 40 is connected to the second input of the differential amplifier 34. The outputs of memories 37 and 38, in which are stored the digital values of the characteristic samples of l image, are connected to the inputs of a multiplexer 41 which supplies the series of characteristic digital image values in the same order as to the output of the differential amplifier, the output of this multiplexer 41 being connected to the input d 'a programmed memory 42 providing processing of the gamma correction, blanking type, etc. The output of this memory 42 is connected to the input of a demultiplexer which supplies digital values characteristic of the three color components, V, X, and B. A time base and management circuit 44 delivers a control signal to the imager 20 for the transfer and the offset in the output charge transfer registers, a control signal to the circuit. t digital level alignment and noise correction with fixed structure 40, control signals to memories 37, 38 and 39, a control signal to programmed memory 42, and control signals to multiplexers 33 and 41 and to the demultiplexers 36 and 43. Finally, this management and time base circuit 44 supplies the Hech sampling clock signal to the analog-digital converter. This clock signal has a frequency twice the offset frequency of the output registers of the imager, that is to say four times the frequency of the color subcarrier, ie 17.76 MHz.

 In general, the operation of these processing circuits is as follows: the signals from the imager 20 are amplified separately by the two pre-amplifiers, the outputs of which are multiplexed to drive a differential amplifier. The speed of this amplifier is twofold since it delivers on the one hand a signal to the analog-digital converter and on the other hand it performs the function of level alignment and noise correction with fixed structure, in the way which will be described below. At the output of the analog-to-digital converter, the video signal in digital form is transmitted to one or the other of the three memories 37, 38 and 39. This switching is carried out as follows: during the frame blanking period the system controls the acquisition of noise with a fixed structure existing between the different columns of the imager matrix. This acquisition of noise with a fixed structure is obtained by reading the content of the different columns.

After transfer to the output charge transfer registers, and sampling, this information is stored in memory 39 known as fixed structure noise memory. This noise reading is performed at the start of each frame. Before the acquisition of the fixed structure noise the signal is aligned thanks to the digital level alignment circuit. Such a digital alignment circuit - of the level of an analog signal is described in French patent application Nb 82 05783 filed April 2, 1982 in the name of the Applicant.

 This alignment circuit 40 comprises an error detection and correction calculation circuit which at each alignment pulse compares a detected digital value with a reference value characteristic of the alignment level, an adder which adds the calculated correction to the previous correction value, and a memory register which stores the correction value in memory until the next alignment pulse. The correction detection and calculation circuit is connected to a digital-analog converter. The noise values stored in memory 39 are applied to an additional input of the adder.

 Thus, the analog values which appear on the first input of the amplifier 34 are combined in this amplifier with a signal characteristic of the alignment value calculated by the digital alignment circuit to which is added information coming from the memory of fixed structure noise 39 This signal combined with the sample applied to the second input of amplifier 34 is characteristic of the offset which exists between the column from which the sample comes and the reference black level. Such a calculation is reproduced with each sample in order to allow the black level of the different columns to be aligned. The samples delivered by the output of the analog-digital converter are therefore free from noise with fixed structure and their values are fixed with respect to a stable reference level.

 These samples are transferred to the two memories 37 and 38 in the following manner. These two memories work in flip-flop, that is to say that we write in one while we read in the other and vice versa. Each memory can store two lines of the video image. This provision is made so that the speed of writing and reading in the memories is compatible with the speed of output of the samples from the analog-digital converter. The outputs of these two memories are connected to a multiplexer 41 which reproduces the series of interleaved samples. The output of this multiplexer is connected to - the input of the programmed memory 42 which performs the functions of gamma correction, blanking and possibly clipping of the black and white levels. The output of the programmed memory 42 is connected at the input of the demultiplexer 43 which delivers the signals R, G, B, distributed in time as indicated in FIG. 9, the green channel comprising a series of samples at the sampling frequency, 8.86 MHz and the red and blue channels comprising sequences of samples at the frequency half of the sampling frequency. The information identified by V and B belong to the current line while the information R 'and B' belong to the next line. The outputs of these three channels are connected to a processing circuit to deduce therefrom the luminance component Y and the color difference components DR and DB.

For this, the green signal at the output of channel B is separated into two components by means of two filters, 47 and 48 respectively low pass and high pass, the cutoff frequency of which is fixed at 750 KHz. The digital low-pass filter is a transverse filter with five weights. The output of the digital high-pass filter 48 is connected to the input of a rejector circuit 49 for which the rejection frequency is 4.4 MHz. The essential role of this filter rejector is to eliminate the effects of serrations. The output signal of this rejector filter is the pseudo-luminance signal YHF. The outputs of the rejector filter 49 and of the digital low-pass filter 47 are connected to the first two inputs of a matrixing circuit 52 which supplies on a first output the luminance component Y = YBF + K YHF. The red channel is connected to the input of a digital low-pass filter 50, the cut-off frequency of which is also fixed at 750 KHz. Likewise, channel B is connected to the input of a digital low-pass filter 51 with the same cut-off frequency FC = 750 KHz. The outputs of filters 50 and 51 are connected to two additional inputs of the matrixing circuit 52 which provides on a second output the color difference component DR = R-YBF, and and the color component
DB = B-YBF. The three signals Y, DR and DB can then be transmitted to a PAL or SECAM digital encoder.

 The camera thus described can of course be the subject of different embodiments linked to the various components which may be used. Depending on the components chosen, it may be necessary to perform certain functions in parallel in several channels which will then be multiplexed, insofar as the operating speeds of the various elements are not sufficient.

However, in the structure as described, such a camera can be produced with a SOS MOS digital-analog converter, and very fast CMOS memories.

 L; FIG. 8 represents another embodiment of the processing circuits of the camera according to the invention in which processing is carried out from analog signal samples, before analog-digital conversion in particular for the alignment of the level of black. The same elements as in FIG. 7 have been designated in this figure by the same references. The outputs Sp and S. of the imager 20 are connected to the inputs of two video alignment circuits, 60 and 61, the outputs of which are connected to the inputs of the multiplexer 33. The output of the multiplexer 33 is directly connected to the input of the analog-to-digital converter 35. The output of this analog-to-digital converter is, as in the previous figure, connected to the inputs of three memories, two line memories 37 and 38 intended for the storage of the digitized samples characteristic of the image, and a memory 39 in which are stored the characteristic values of the noise with fixed structure in the different columns of the matrix. The outputs of the memories 37 and 38 are connected to the multiplexer 41, the output of which is connected to the programmed memory 42 not directly as in FIG. 7, but via a fixed structure noise suppression circuit 63. This fixed structure noise suppression circuit, which can be a simple subtractor, has a second input connected to the output of the fixed structure noise memory 39. The series of digital samples present at the output of the suppression circuit 63 therefore comprises noise-free samples with a fixed structure and whose levels are fixed relative to a stable reference, the processing having been carried out upstream. The output of the fixed structure noise suppression circuit is connected to the input of the programmed memory 42 which performs the gamma correction and the blanking correction. The output of this memory is connected to the input of the demultiplexing circuit 434 The other circuits are similar to those described with reference to FIG. 7. A management and time base circuit 64 has an output intended for controlling the 2Q imager, outputs connected to the control inputs of the video alignment circuits 60 and 61, an output connected to the control input of the fixed structure noise memory 39, an output connected to the control input of the circuit noise cancellation with fixed structure 63 and two additional outputs connected respectively to the control inputs of the program memory 42 and of the demultiplexing circuit 43. As can be seen, this embodiment includes a larger analog processing part than the previous figure since the leveling of the blacks is done analogically.

 The monosensor color camera thus constructed has a fairly simple structure and can be constructed from elements of average performance while guaranteeing quite satisfactory image quality.

Claims (5)

 1. Monosensor color television camera, characterized in that it comprises in combination:  - a solid state imager (20) with line transfer having two outputs (Si, Sp) delivering signals associated respectively with the points.  - and circuits for processing the output signals from the imager to form the three components of a color TV image, the formation of this image being obtained by controlling the reading of all the lines of the imager successively for the duration d a frame of the image, two successive lines being read for the duration of a TV line and combined in the processing circuits to form the components of a line of the TV image.  a chromatic filter (10) composed of elementary filters associated with photosites such that the filters of the same rank in the successive lines of a frame of the matrix are similar, an alternation of green and blue filters being carried out on the lines of a screen and an alternation of red and green filters being carried out on the lines of the other screen, the green filters being staggered over the whole of the matrix, odd and even points of the successive lines of the imager photosite matrix,  2. Camera according to claim 1, characterized in that two successive lines of the matrix of rows n and n ', n and n' being the current indices of the lines of the two frames of the photosite matrix, are combined to form - a line of a first frame of the image, the successive lines of rows n ′ and n + 1 being combined to form a line of the other frame of the image, the two fields of the image thus being interleaved.  3. Camera according to claim 1, characterized in that the processing circuits comprise an analog-digital converter (35) whose output is coupled on the one hand to a first memory (37, 38) for the storage of the values digital characteristics of the image and on the other hand to a second memory for the storage of digital values characteristic of the noise with fixed structure (39), a noise suppression circuit (63) whose inputs are coupled to the outputs of the two memories, and a programmed memory (42) for gamma correction which processes the series of noise-free digital values.  4. Camera according to claim 1, characterized in that the processing circuits include an analog-digital converter (35), the output of which is coupled to a memory (37, 38) - for storing digital values characteristics of the image, and on the other hand to a first input of a level alignment circuit (40) and to the input of a so-called fixed-structure noise memory (39) whose output is coupled at a second input of the alignment circuit (40), this alignment circuit determining for each analog signal transmitted by the imager to the converter the reference level to be considered for determining the digital value, and a programmed memory (42 ) for gamma correction which processes the sequence of noise-free digital values.  5. Camera according to one of claims 3 and 4, characterized in that the output of the programmed memory (42) is coupled to the input of a demultiplexer (43) with three outputs, to which the characteristic digital values are transmitted components of respectively green, red and blue colors, which are connected to filtering circuits (47-51), the outputs of the filtering circuits being connected to the inputs of a matrixing circuit (52) which delivers the components of the TV picture.