JPH05191799A - Digital picture compressor with film picture detection function - Google Patents

Abstract

PURPOSE:To limit number of quantization bits, to allow picture quality between frames to hardly cause a difference and to improve the transmission efficiency by discriminating a type of an input picture signal. CONSTITUTION:A subtractor 2, a DCT circuit 3, a quantization circuit 4, an inverse quantization circuit 5, an adder 8, a frame memory 10, a motion compensation circuit 9, a loop filter 7, a motion detection circuit 11 quantize a picture signal and a variable length coding circuit 12 implements the variable length coding. In this system, in-frame processing and inter-frame processing are selected adaptively in response to a picture motion of the input picture signal and its output is inputted to a buffer memory 13. In this case, the periodicity of a memory occupancy rate of data in the buffer memory 13 is detected by a periodicity detection circuit 21 and when any periodicity is in existence, a quantization table selection circuit 22 limits a bit number of quantization to be a prescribed value or below.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital image compression apparatus, and more particularly to a digital image compression apparatus with a film image detection function, which detects a film image when it is input and performs compression suitable for the film image.

[0002]

2. Description of the Related Art In recent years, various types of television systems have been developed in response to the need to enjoy a wide variety of images. However, in the current broadcasting system, the number of images that can be transmitted within a predetermined band is limited, and it is the current situation that it cannot meet various needs. One of the causes is that the current broadcast has a fixed band because it is an analog signal transmission and is not suitable for image compression transmission. Therefore, recently, techniques for digital transmission of images and band compression have been developed. By using these techniques, it is of course possible to transmit a wider variety of images than in the current broadcasting, but it is also possible to transmit a wider variety of images by increasing the image compression rate.

However, current analog broadcasting, for example NTS
In the case of the C system signal, if the analog signal is simply converted into a digital signal, the rate will be about 100 Mbps, so 2
Highly efficient image compression of 0 to 50 times is required. Note that N
The TSC signal is generally a component television signal of 525 scanning lines.

There are various image compression methods, but one of the most efficient methods is a conversion code for reducing the spatial redundancy of inter-frame prediction and prediction error as shown in H261 of CCITT recommendation. There is a hybrid coding system that combines the coding and the coding. FIG. 5 shows an image compression device.

An image signal is input to the input terminal 1 and supplied to the subtractor 2 and the motion detection circuit 11. The subtracter 2 performs a subtraction process described later, and the output of the subtractor 2 is input to the discrete cosine transform (hereinafter referred to as DCT) circuit 3. D
The CT circuit 3 takes in 8 pixels in the horizontal direction and 8 pixels in the vertical direction as a unit block (8 × 8 pixels = 64 pixels), and outputs a coefficient obtained by converting the pixel array from the time axis domain to the DCT transform domain. Here, each coefficient is input to the quantization circuit 4,
Is quantized. The quantization circuit 4 has 16 or 32 types of quantization tables, and each coefficient is quantized based on the selected quantization table. The quantizing circuit 4 is provided with a quantizing table so that the amount of generated information and the amount of transmitted information are within a certain range. The coefficient data output from the quantization circuit 4 is zigzag-scanned from the low frequency band to the high frequency band for each unit block, is extracted, and is input to the variable length encoding circuit 12.
Variable length coding is performed with a set of a number (run length) following zero coefficients and a non-zero coefficient. The encoder is a variable length encoder having a different code length depending on the frequency of occurrence of Huffman code or the like. The variable length coded data is stored in the buffer memory (FIFO).
Circuit 13 to be read out at a prescribed speed, supplied to a multiplexer in the next stage (which multiplexes a control signal, voice data, synchronization data (SYNC), etc.) and sent out to a transmission line. Although data is sent to the buffer memory 13 at a variable transfer rate, this data is temporarily written in the buffer memory 13 and is output at a fixed rate necessary for channel transmission.

The output of the quantization circuit 4 is input to the inverse quantization circuit 5 and inversely quantized. Further, the output of the inverse quantization circuit 5 is input to the inverse DCT circuit 6 and returned to the original signal. This signal is input to the frame memory 10 via the adder 8. The output of the frame memory 10 is supplied to the motion compensation circuit 9 and the motion detection circuit 11. The motion detection circuit 11 compares the input signal from the input terminal 1 with the output signal of the frame memory 8 to detect a motion vector of an image block unit (for example, 8 × 8 pixels or 16 × 16 pixels) and perform motion compensation. Output to the circuit 9. In the motion compensation circuit 9,
The signal input from the frame memory is compensated according to the detected motion vector amount so that the original image matches the image one frame before. However, if the motion vector amount exceeds the compensable range, the output is cut off. The output of the motion compensation circuit 9 is passed through the loop filter 7 to the subtractor 2
Is supplied to. The output of the motion compensation circuit 9 can also be fed back to the adder 8 and the frame delay unit 8 via the loop filter 7. Next, the basic operation of the above system will be described. The basic operation of this system is
There are (1) intra-frame coding processing and (2) inter-frame coding processing.

The intra-frame coding process is performed as follows. The input image signal that has passed through the input terminal 1 and the subtractor 2 is converted from the time domain to the DCT domain by the DCT circuit 3 and quantized by the quantization circuit 4. The quantized signal is output to the transmission path via the buffer memory 13 after undergoing variable length coding processing. The quantized signal is returned to the original signal by the inverse quantization circuit 5 and the inverse DCT circuit 6, and delayed by the frame memory 10. Therefore, in the intra-frame coding process, it is equivalent to that the information of the input image signal is variable-length coded as it is. This intra-frame coding process is performed when the scene of the input image signal changes or when the motion vector amount exceeds the compensation amount of the motion compensating circuit, and further in an appropriate cycle in a predetermined block unit.

Next, the interframe coding process will be described. When the interframe coding process is executed, the subtractor 2 obtains a signal corresponding to the difference between the input image signal and the image signal one frame before. This difference signal is input to the DCT circuit 3, and D from the time domain
It is converted into the CT conversion domain and then quantized by the quantization circuit 4. Further, since the differential signal and the image signal are added by the adder 8 and input to the frame delay unit 8, a predicted image signal that predicts the input image signal that is the source of the differential signal is created and input. Will be.

In the case of the inter-frame coding process, the difference information between one frame before and the original signal is coded and transmitted. Image information can be transmitted with a high compression rate for this purpose. The part surrounded by the broken line in the figure is a part called a local decoder.

At the receiving side, it is determined, for example, from the control data transmitted at the same time whether the signal subjected to the intra-frame coding process has arrived or the signal subjected to the inter-frame coding process has arrived. .. When a signal for which intra-frame coding processing has been performed arrives, inverse quantization, inverse D
CT may be performed. When a signal subjected to interframe coding processing arrives, the signal one frame before and the arrived signal may be added. Here, considering the type of image signal input to the input terminal 1, an input image signal from a general camera and a film image signal can be considered. There is no problem in processing an input image signal from a camera, but in the case of inputting an image signal made from film, 3-2 pulldown processing is performed in advance as described below.

FIG. 6 (a) shows a film image,
One frame of this film image is 24 / sec. On the other hand, FIG. 6B shows an NTSC non-interlaced image. In this image, one frame is 1/60 second.
Therefore, to convert a film image into an NTSC image, as shown by the arrow in FIG.
If you use it 2 times, 3 times, and so on, NT
It can be converted to SC image. Here, if the image signal converted as described above is input to the image compression apparatus described with reference to FIG. 5 and processed, the following problems occur.

FIG. 7 shows a change state of information in the image compression apparatus. FIG. 7A shows a state of generation of the information amount when the film image signal is input. In the case of a film image signal, as described above, since the 3-2 pulldown process is performed, the amount of information generated in 2 out of 5 frames is concentrated. this is,
This is because the same image data as 3 frames, 2 frames, 3 frames, ... Is input and the subtractor 2 obtains the difference from the image signal of the previous frame.

When the amount of generated information largely changes as described above, the occupied amount of the buffer memory 13 also largely changes with time as shown in FIG. 7 (c). Then, when the control for determining the number of quantization bits is performed by monitoring the data occupation amount of the buffer memory 13, a frame with almost no information amount as shown in FIG. 7B (A3 in the example of FIG. 7A) is obtained. Frame), the number of quantization bits will be increased more than necessary. As a result, there is a problem that the difference in image quality increases depending on the frame.

[0014]

As described above, according to the conventional image compression apparatus, when the input image signals having the same image content are input with a periodic regularity, the quantization bit number switching control is performed. , The number of quantization bits may increase even for an input image signal with a small amount of information,
The difference in image quality from the image signals of other frames may be noticeable.

Therefore, the present invention provides a digital image compression apparatus with a film image detection function, in which the number of quantization bits is limited by discriminating the type of an input image signal, and the difference in image quality between frames is less likely to occur. The purpose is to do.

[0016]

SUMMARY OF THE INVENTION According to the present invention, there are provided information obtained by performing image compression coding processing on an input image signal using information within a frame and information subjected to image compression coding processing using information between frames. , Adaptively switching and outputting according to the image movement of the input image signal, temporarily storing output information in a buffer memory, and monitoring the amount of information in this buffer memory to perform quantization in the image compression encoding processing. In the digital image compression apparatus having means for controlling the table, when the periodicity of the information amount of the buffer memory is detected and it is detected that the input image signal has a predetermined periodicity, the quantization table is changed. It comprises means for selecting and forcibly fixing the quantization bit range within a predetermined range.

[0017]

According to the above-mentioned means, even in the case of a film image in which the bit range of quantization is largely changed in the normal image compression coding processing, the type of the input is detected and the quantum within the predetermined range is automatically detected. Since it is fixed to the number of encoded bits, there is no big difference in image quality.

[0018]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows an embodiment of the present invention. Since the basic configuration is the same as that of the device shown in FIG. 5, common parts will be described with the same reference numerals.

The part different from the conventional device shown in FIG. 5 will be described below. That is, the buffer memory 1 for transmission
That is, the signal indicating the information amount of 3 is supplied to the periodicity detection circuit 21 and the quantization table selection circuit 22. Further, the periodicity detection signal of the periodicity detection circuit 21 is supplied to the quantization table selection circuit 22. When the signal indicating the periodicity is not obtained from the periodicity detection circuit 21, the quantization table selection circuit 22 regards the currently processed input image signal as a normal image signal, and the quantization table selection circuit 22 performs the quantization as in the conventional case. The quantization table of the digitizing circuit 4 is controlled.

However, when the signal of which the periodicity is detected is obtained from the periodicity detection circuit 21, the selection mode of the quantization table is switched to forcibly limit the upper limit of the number of quantization bits.

In the periodicity detection circuit 21, if the threshold level TH is set for the information indicating the occupied amount of the buffer memory as shown in FIG. Can be determined.

FIG. 2A shows a concrete example of the configuration of the periodicity detection circuit 21. The occupancy rate in the buffer memory 13 can be known from the difference between the values of the word length count accumulator 211 of the input image signal and the word length count accumulator 212 of the output image signal. This difference is subtracted from the subtracter 213
And is supplied to the quantization table selection circuit 22. When the difference is large, the occupancy rate is large, and when the difference is small, the occupancy rate is small. Therefore, by monitoring the change in this difference value,
It is possible to determine whether or not the 3-2 pulldown image signal is input. When the difference value changes periodically, the quantization table selection circuit 22 determines this by setting a threshold level, for example.

When a 3-2 pull-down image signal is input, the quantization table selection circuit 22 operates, for example, as shown in FIG.
A table selection characteristic B as shown in (B) is set. The horizontal axis of FIG. 2B is the occupation rate information of the buffer memory 13, while the vertical axis is the quantization bit amount. When the occupancy rate is small, the number of quantization bits is increased to make it dense, and when the occupancy rate is large, the number of quantization bits is decreased to make it coarse. The selection characteristic (A) is a characteristic when a normal image signal is input, while the selection characteristic (B) is a characteristic when a 3-2 pulldown image signal is input. In this way, when quantization is performed by limiting the occupancy rate so that it does not exceed a certain range, even if the input image signal has a large amount of information, the number of quantization bits is controlled to be smaller than usual. It will be. This is due to the resolution due to quantization compared to an image signal with a small amount of information,
It means that there is no significant difference, that is, there is no difference in image quality between frames. It also means that the amount of transmitted information will be more efficient than before.

The present invention is not limited to the above embodiment. In the above embodiment, the periodicity of the occupancy of the buffer memory 13 is judged, the quantization table is selectively controlled, and the number of quantization bits is limited. In addition to this function, the detection information from the periodicity detection circuit 21 may be further given to the header insertion circuit 23 and transmitted as header information. By doing so, the receiving side can easily determine that the compressed signal of the 3-2 pulldown image signal is received by determining the header information. The result of determining the header information can be used for various purposes. For example, it is a display or the like indicating that the 3-2 pulldown image signal is being received.

FIG. 3 shows another embodiment of the present invention. In the embodiment of FIG. 1, it is detected from the periodicity of the data occupation rate of the buffer memory 13 that the 3-2 pulldown image signal is input. However, as a method of detecting the periodicity, there is also a method of detecting from the selection information of the quantization table selection circuit 22. Quantization table selection circuit 2
2 selects the quantization table of the quantization circuit 5, and the selection information is input to the periodicity detection circuit 25. When the selection information has a periodicity, the periodicity detection circuit 25 can feed back the periodicity detection information to the quantization table selection circuit 22 to realize the limitation of the number of quantization bits. Also in this embodiment, the periodicity detection information may be transmitted to the receiving side as header information.

In the above-described embodiment, the periodicity is judged by detecting that the data occupation rate of the buffer memory has increased. However, conversely, it is possible to determine the periodicity by detecting a frame in which no information amount occurs. When an image signal obtained by 3-2 pulling down a film image into an interlaced image is image-compressed, the coarse and dense transition of the information amount shows that the information amount in the shaded field in FIG. 4 is always coarse. In other words, once every 5 fields, the amount of information is coarse. Therefore, this periodicity is defined as the occupancy rate of the buffer memory,
Alternatively, the table selection information of the quantization table or the like may be detected to determine that the image signal is a 3-2 pulldown image signal.

[0027]

As described above, according to the present invention,
By determining the type of the input image signal, it is possible to limit the number of quantization bits, prevent a difference in image quality between frames from occurring, and improve the transmission efficiency.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention.

FIG. 2 is a specific circuit diagram of the cycle detection circuit of FIG. 1 and a characteristic diagram of a quantization table selection circuit.

FIG. 3 is a block diagram showing another embodiment of the present invention.

FIG. 4 is an explanatory diagram of an image signal obtained by 3-2 pulling down a film image shown for explaining an operation example of the apparatus of the present invention into an interlaced image.

FIG. 5 is a block diagram showing a conventional image compression device.

FIG. 6 is an explanatory diagram of an image signal in which a film image is 3-2 pulled down to an interlaced image.

FIG. 7 is an explanatory diagram of a signal quantization amount and a buffer memory occupation amount by a conventional image compression apparatus.

[Explanation of symbols]

2 ... Subtractor, 3 ... DCT circuit, 4 ... Quantization circuit, 5 ... Inverse quantization circuit, 6 ... Inverse DCT circuit, 7 ... Loop filter,
8 ... Adder, 9 ... Motion compensation circuit, 10 ... Frame memory, 11 ... Motion detection circuit, 12 ... Variable length coding circuit, 1
3 ... Buffer memory, 21 ... Periodicity detection circuit, 22 ... Quantization table selection circuit.

Claims (4)

[Claims] 1. An image motion of the input image signal, which includes information obtained by performing image compression coding processing on the input image signal using information within a frame and information obtained by image compression coding processing using information between the frames. A unit for controlling the quantization table in the image compression encoding processing by adaptively switching and outputting according to the above, temporarily storing output information in the buffer memory, and monitoring the amount of information in this buffer memory In the digital image compression device, a periodicity detection unit that detects the periodicity of the information amount change of the image-compressed signal, and when the periodicity detection signal is obtained from the periodicity detection unit, the quantization table is used. A digitizer with a film image detecting function, comprising: a quantization table selecting means for forcibly fixing a quantization bit range within a predetermined range. Image compression device. 2. The digital image compression apparatus with a film image detecting function according to claim 1, further comprising a sending means for sending the periodicity detection signal of said periodicity detecting means to a receiving side. 3. The periodicity detecting means detects the periodicity of the change in the information amount by detecting a large change in the data occupancy of the buffer memory.
A digital image compression device with the film image detection function described. 4. The periodicity detecting means detects the periodicity of the change in the information amount by detecting the change in the control output of the means for controlling the quantization table. Digital image compression device with film image detection function.