JPH06268993A - Interlaced scanning moving image encoding device - Google Patents

Abstract

PURPOSE:To provide a blocking system optimum for motion compensation threshold value judgement by utilizing the property of orthogonal transformation encoding. CONSTITUTION:Threshold value judgement is performed corresponding to a threshold value judgement offset value command 110 in an intra/inter field macroblocking processor 2 for prediction error frame pictures 102 for which prediction frame pictures 108 are subtracted from present frame pictures 101 by a subtractor 1 and divided intra or inter field macroblocking pictures 103 are orthogonally transformed and quantized by an orthogonal transformation device 3 and a quantizer 4. By using the motion compensated result information 109 of a frame or macroblock unit generated by performing optimum motion compensation by pattern matching with the present frame pictures 101 by a motion compensation prediction device 8 for local decoding frame pictures for which a quantization coefficient 105 is decoded and generated in an inverse process by an inverse quantizer 5, an inverse orthogonal transformation device and an adder 7, a threshold value judgement offset value is calculated by the frame or macroblock unit of the prediction error frame pictures 102 by a threshold value judgement offset value deciding device 9 and the command 110 is issued.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an interlaced scanning moving image encoding apparatus for highly efficiently compressing 2: 1 interlaced scanning moving images.

[0002]

2. Description of the Related Art A conventional interlaced-scan moving image encoding apparatus disclosed in, for example, Japanese Unexamined Patent Publication No. 3-1688, as shown in FIG. 4, has a macroblocking processor 2a for a 2: 1 interlaced scanning current frame. The image 101 is divided into a macroblocking image (black and white circles represent first and second field pixels) 114 of 8 pixels × 16 lines as shown in FIG. 5A, for example. The sub-block input / output switches 12 and 13 are
Macro motion determination command 113 from macro motion determiner 15
Accordingly, the macroblocking processor 2a is switched to connect the macroblocking image 103a to one of the intra-field / interfield subblocking processors 2b and 2c. Further, the intra-field / inter-field sub-blocking processors 2b and 2c provide the intra-field / inter-field sub-blocking image 103.
Either b or 103c is switched to be connected to the discrete cosine converter 3a. The intra-field / inter-field sub-blocking processors 2b and 2c are the sub-blocking input switch 12
The macroblocking image 103a from FIG.
As shown in (b), it is divided into an intra-field sub-blocking image 103b of 8 pixels × 8 lines, which is composed of only pixels of odd and even lines. Also, for example, in FIG.
As shown in (c), it is divided into an inter-field sub-blocking image 103c of 8 pixels × 8 lines, which is composed only of pixels of 1 to 8 lines and 9 to 16 lines. The discrete cosine transformer 3a performs a discrete cosine transform (DCT) on one of the intra-field / inter-field sub-blocking images 103b and 103c from the sub-blocking output switch 13 to obtain 8 × 8 DCs.
The T coefficient 104a is generated. The low frequency / residual component quantizers 4a and 4b, the low frequency / residual component encoders 5a and 5b, and the adder 14 are 4 × 4 of the 8 × 8 DCT coefficients 104a from the discrete cosine transformer 3a. Number of low frequency components and (8 × 8) − (4 × 4) residual component components are quantized and encoded, and then added to obtain encoded data 11
Generates 2. The macro motion determiner 15 determines from the macroblocking processor 2a whether the macroblocking image 103a is moving or stationary, and issues a motion / stillness determination command 113.

The interlaced-scan moving image encoding apparatus of the above-mentioned conventional example is a method of performing macro-motion determination on a macro-blocking image 103a to be encoded and performing intra-field or inter-field sub-blocking encoding (macro-motion determination sub-blocking encoding). Method).

[0004]

In the conventional interlaced scanning moving picture coding apparatus as described above, the reason why the macro motion judgment is effective for judging the necessity of the intra-frame and inter-frame sub-blocking coding is as follows. , Statistically the moving area is in the field,
The static region has a high correlation between fields. When performing orthogonal transformation on a prediction error frame image using motion-compensated prediction in a normal high-efficiency moving image coding method, there is a problem that the method of realizing the motion / stillness determination is not clear and the coding efficiency may deteriorate. It was

The problem to be solved by the present invention is to provide a method of performing intra-field or inter-field macro-blocking coding by performing motion compensation threshold value judgment on a prediction error frame image to be coded by an interlaced scanning video image coding device (motion coding). Compensation threshold determination macroblocking coding method).

[0006]

In order to solve the above-mentioned problems, the interlaced scanning moving image coding apparatus of the present invention is provided with the following means, and adopts a motion compensation threshold value judgment macroblocking coding system. .

The subtractor subtracts the current frame image of the 2: 1 interlaced scan from the prediction frame image from the motion compensation predictor to generate a prediction error frame image.

The intra-field / inter-field macroblocking processor makes a threshold judgment in field units for the prediction error frame image from the subtractor according to the threshold judgment offset value command from the threshold judgment offset value determiner, and determines M pixels × M pixels. Divide into macroblocking images within or between N line fields.

The orthogonal transformer and quantizer are
The transform coefficient obtained by performing the orthogonal transform operation on the intra-field or inter-macro blocking image from the inter macroblocking processor is quantized by feedback control from the subsequent stage to generate a quantized coefficient. Alternatively, the quantizer directly inputs the motion compensation result information from the motion compensation predictor as feedforward information from the previous stage, and performs quantization control instead of or in combination with the feedback information from the subsequent stage.

The inverse quantizer, the inverse orthogonal transformer, and the adder use a motion-compensated predictor to generate a decoded prediction error frame image obtained by subjecting the quantized coefficient from the quantizer to the inverse process of quantization and orthogonal transform operation. To generate a locally decoded frame image.

The motion compensation predictor performs a pattern matching operation between the current frame image and the preceding locally decoded frame image from the adder to generate an optimum predicted frame image. In addition, motion compensation statistical information indicating the number of blocks in which intra-field and inter-frame motion compensation is selected on a frame-by-frame or macro-block basis and motion compensation type information indicating whether intra-field or inter-frame motion compensation is selected for the block. It is generated as compensation result information.

The threshold determination offset determiner uses the motion compensation result information regarding the current frame from the motion compensation predictor, and selects the inter-field and intra-motion compensation for each frame or macroblock of the prediction error frame image from the subtractor. A threshold judgment offset value is calculated and instructed from a function for the difference value of the number of blocks. Alternatively, each threshold judgment offset value is calculated from the motion compensation predictor using each motion compensation result information regarding the macroblock and a plurality of peripheral macroblocks, and, for example, each arithmetic mean value is set as the threshold judgment offset value. Alternatively, when the image deterioration of the entire current frame image is large, instead of using the motion compensation result information on the current frame from the motion compensation predictor, the motion compensation result information on the preceding frame that could be coded efficiently in the past several frames is used. A threshold value judgment offset value is calculated.

The motion compensation decision unit decides whether or not motion compensation is necessary depending on whether a difference value between the current frame image and the locally decoded frame image from the adder is larger than a predetermined threshold value. At this time, the locally decoded frame image is output. It also issues an encoding mode identification command for intra-frame direct or inter-frame motion compensation prediction.

The coding mode switching unit, in accordance with the coding mode identification command from the motion compensation judging unit, outputs the current frame image or the locally decoded frame image from the motion compensation judging unit to the intra-field / inter-field macroblocking processor or the motion compensation prediction. Switch to connect to the vessel.

[0015]

The interlaced scanning moving image encoding apparatus of the present invention is the above-mentioned means, and when the deterioration of the image quality of the current frame or the entire current frame image is large with respect to the prediction error frame image to be encoded, the past several frames before. Using the motion compensation result information regarding the preceding frame that can be efficiently coded, the threshold determination offset value is calculated for each frame. Alternatively, for example, an arithmetic mean value of the threshold determination offset values in macroblock units is calculated using the motion compensation result information on the macroblock and a plurality of peripheral macroblocks, and the calculated threshold average offset value is set. Next, according to the threshold value determination offset value, a threshold value determination is performed in field units for the prediction error frame image to be encoded, and M pixels ×
Divide into macroblocking images within or between N line fields. Further, the intra-field or inter-field macroblocking image is orthogonally transformed, and the transform coefficient is fed back from the former stage by feedback control by the quantization characteristic signal from the coding control unit and the motion compensation result information from the motion compensation predictor. Either or both of the controls determine the quantization step size and perform quantization. On the other hand, pattern matching is performed between the current frame image and the preceding locally decoded frame image to generate motion compensation result (statistics and type) information in frame units or macroblock units. Alternatively, whether or not motion compensation is necessary is determined by whether or not the difference value between the current frame image and the preceding locally decoded frame image is larger than a predetermined threshold value, and macroblock coding of either intra-frame direct or inter-frame motion compensation prediction is applied. To choose.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS As shown in FIG. 1 (excluding a code assigning unit and a coding control unit), a subtracter 1 is a 2: 1 interlaced scanning moving image coding apparatus according to an embodiment of the present invention. The current frame image 101 of the scan is subtracted from the prediction frame image 108 from the motion compensation predictor 8 to generate a prediction error frame image 102. In-field / inter-field macroblocking processor 2
According to the threshold value determination offset value command 110 from the threshold value determination offset value determiner 9, the subtracter 1 performs threshold value determination in field units on the prediction error frame image 102,
For example, it is divided into macroblocking images 103 in or between fields of 8 pixels × 16 lines. Orthogonal transformer 3
And the quantizer 4 quantizes the intra-field / inter-field macroblocking processor 2 by applying a transform coefficient 104 obtained by performing an orthogonal transform (for example, DCT) operation to the intra-field or inter-field macroblocking image 103 to the generated information amount. Quantize by feedback control from the latter stage that determines the step width,
The quantization coefficient 105 is generated. The inverse quantizer 5, the inverse orthogonal transformer 6, and the adder 7 include a quantization coefficient 10 from the quantizer 4.
The decoded prediction error frame image 106 obtained by performing the inverse process of the quantization and the orthogonal transformation operation on 5 is added to the prediction frame image 108 from the motion compensation predictor 8 to generate a locally decoded frame image 107. The motion compensation predictor 8 collects the same field or different fields between the current frame image 101 and the preceding locally decoded frame image 107 from the adder 8 as shown in FIG. 5B or 5C, for example. An optimum prediction frame is obtained by configuring a prediction block in or between fields, performing pattern matching calculation, and performing motion compensation in or between fields to obtain one or more motion vectors for each macroblock of 8 pixels × 16 lines. Image 108 is generated. In addition, motion compensation statistical information indicating the number of blocks (macroblocks or sub-blocks) for which intra-field and inter-frame motion compensation has been selected in frame units or macroblock units, and whether intra-field or inter-frame motion compensation has been selected for the block. The motion compensation type information represented is generated as motion compensation result information 109. The threshold value determination offset value determiner 9 follows the motion compensation result information 109 from the motion compensation predictor 8 and determines the number BLK of macroblocks for which inter-field and intra-motion compensation has been selected for each frame of the prediction error frame image 102 from the subtractor 1. Difference value N between _INTER and BLK _INNER (BLK
_INTER- BLK _INNER ), the threshold judgment offset value OS is calculated, and a threshold judgment offset value command 110 for judging the necessity of intra-frame and inter-macro macroblock coding is issued.

The interlaced-scan moving image encoding apparatus of the above-described embodiment is a method of performing motion compensation threshold value determination on the prediction error frame image 102 to be encoded and intra-field or inter-field macroblocking encoding (motion compensation threshold value determination macro). Blocking coding method) is adopted.

The intra-field / inter-field macroblocking processor 2 follows the threshold value judgment offset value command 110 from the threshold value judgment offset value deciding device 10 and subtracts the prediction error frame image 102 from the subtracter 1 in accordance with the threshold judgment formula in a field unit. When ER _INTER <ER _INNER + OS For example, a field blocking image 103 of 8 pixels × 16 lines,
When ER _INTER ≧ ER _INNER + OS, for example, 8 pixels ×
Each is divided into 16 lines of inter-field blocking image 103. If the threshold determination offset value OS is large, the number of inter-field macro blocking images 103 increases. ER _INTER = Σ _PIXEL Σ _{LINE │I} (PIXEL, LINE) -I (PIXEL, LINE + 1) │ PIXEL = 1 to 8, LINE = 1, 3, 5, ..., 15 ER _INNER = Σ _PIXEL Σ _LINE {| I (PIXEL, LINE) -I (PIXEL, LINE + 2) | + | I (PIXEL, LINE + 1) -I (PIXEL, LINE + 3) |} PIXEL = 1 to 8, LINE = 1, 5, 9, ... 16 Here, ER _INTER and ER _INNER are sums of absolute differences between pixel values in macroblocks between fields and I, PIXE
L, LINE) is the pixel position PIXEL and the line position LI
It represents a pixel value (usually a luminance value) in the macroblock on the NE.

As shown in FIG. 2, the threshold value judgment offset value determiner 9 is a completely linear type (FIG. 2 () in response to the difference value N (BLK _INTER −BLK _INNER ) from the subtracter 1 for each frame of the prediction error frame image 102. a)), section linear type (see FIG. 2B), and step type (see FIG. 2C) functions, the threshold determination offset value OS is calculated, and a desired threshold determination offset value command 110 is calculated. To do.

Although the above embodiment has been described assuming that the inter-frame motion compensation predictive coding system is adopted, the motion compensation determining unit 10 and the coding mode switching unit 11 are provided as shown in FIG. It is also possible to judge and adaptively select the intra-frame direct encoding method for independently encoding the current frame image 101 without referring to the locally decoded frame image 107. Good results are obtained with frame images where motion compensation does not work effectively. The motion compensation determiner 10 determines whether or not motion compensation is necessary depending on whether the difference value (motion compensation prediction error value) between the locally decoded frame image 107 from the adder 7 and the current frame image 101 is larger than a predetermined threshold. When it is large, the current frame image 101 is output, and when it is small, the locally decoded frame image 107 is output. At the same time, a coding mode identification command 111 for identifying each coding mode of intra-frame direct and inter-frame motion compensation prediction is issued. The coding mode switching device 11 follows the coding mode identification command 111 from the motion compensation determination device 10 to send the current frame image 101 or the locally decoded frame image 1 from the motion compensation determination device 11.
07 is switched to be connected to the intra-field / inter-field macroblocking processor 2 or the motion compensation predictor 8.

In the above embodiment, the quantizer 4 has been described as adopting the feedback quantization control method from the latter stage, but the motion compensation result information 109 from the motion compensation predictor 8 is directly input and the feed from the previous stage is performed. A forward quantization control method may be used. It goes without saying that they may be used in combination.

In the above embodiment, the threshold value determination offset value determiner 9 is described as calculating the threshold value determination offset value for each frame of the prediction error frame image 102. The threshold determination offset value may be, for example, an arithmetic mean value of the threshold determination offset values in macroblock units calculated using the compensation result information 109. When there is a large change in a part of the macroblocks in the current frame image 101, it is possible to absorb a defect in which the difference in the decoded image from the surrounding macroblocks is noticeable. Further, when the image quality deterioration of the entire current frame image 101 is large, instead of using the motion compensation result information regarding the current frame, the motion compensation result information regarding the preceding frame that can be efficiently encoded in the past several frames is used, and the threshold value for each frame. The determination offset value may be calculated. It is possible to realize efficient blocking coding that is not affected by sudden image quality deterioration.

[0023]

As described above, the interlaced scanning moving picture coding apparatus according to the present invention employs a method of performing intra-field or inter-field macro-blocking coding on the prediction error frame image to be coded. Therefore, the optimal blocking is determined between pixels with high correlation as compared with the conventional method of performing macro motion determination on the macroblocking image to be encoded and performing intra-field or inter-subblocking encoding. It is possible to realize advantageous high-efficiency compression during orthogonal transformation. In addition, the motion compensation threshold value determination can be performed on a frame-by-frame basis or on a macroblock-by-macroblock basis, and there is an effect that it is possible to provide a general-purpose and efficient encoding method that is not affected by the characteristics of the image and adapted to the motion of the image.

[Brief description of drawings]

FIG. 1 is a functional block diagram of an interlaced scanning moving image encoding device according to an embodiment of the present invention.

FIG. 2 is a diagram illustrating a function of a threshold value determination offset value determiner shown in FIG.

FIG. 3 is a functional block diagram of another embodiment shown in the present invention.

FIG. 4 is a functional block diagram of a conventional interlaced scanning moving image encoding device.

5 is a diagram for explaining the functions of the macroblocking processor and the intra-field and inter-subblocking processor shown in FIG.

[Explanation of symbols]

 1 Subtractor 2 In-field / inter-field macroblocking processor 3 Orthogonal transformer 4 Quantizer 5 Inverse quantizer 6 Inverse orthogonal transformer 7 Adder 8 Motion compensation predictor 9 Threshold judgment offset value determiner 10 Motion compensation judgment Device 11 Coding mode switcher 101 Current frame image 102 Prediction error frame image 103 In-field or inter-field macroblocking image 104 Transform coefficient 105 Quantization coefficient 106 Code prediction error frame image 107 Local decoding frame image 108 Prediction frame image 109 Motion compensation result Information 110 Threshold value judgment offset value command 111 Coding mode identification command In the drawings, the same reference numerals indicate the same or corresponding parts.

Claims (5)

[Claims] 1. A subtracter that subtracts a current frame image of 2: 1 interlaced scanning from a prediction frame image from a motion compensation predictor to generate a prediction error frame image, and a threshold determination offset from a threshold determination offset value determiner. According to the value command, the subtracter determines the threshold value for each field for the prediction error frame image, and determines M pixels × N lines (M and N are positive integers).
Intra-field / inter-field macroblocking processor for dividing into intra-field or inter-field macroblocking images, and transform coefficients obtained by performing an orthogonal transform operation on the intra-field or inter-field macroblocking image from the intra-field / interfield macroblocking processor An orthogonal transformer and a quantizer which quantize the quantized signal by feedback control from a subsequent stage and generate a quantized coefficient, and a decoding prediction error in which the quantized coefficient from the quantizer is subjected to an inverse process of quantization and orthogonal transform operation. An inverse quantizer, an inverse orthogonal transformer, and an adder that add a frame image to a predicted frame image from the motion-compensated predictor to generate a locally decoded frame image, and a preceding local decoding from the current frame image and the adder. Pattern matching calculation is performed with the frame image to generate the optimum predicted frame image and Is motion compensation result information indicating motion compensation statistical information indicating the number of blocks for which intra-field / inter-frame motion compensation is selected in macroblock units and motion compensation type information indicating whether intra-field or inter-frame motion compensation has been selected for the block. The number of blocks for which inter-field and intra-frame motion compensation is selected for each frame of the prediction error frame image from the subtracter using the motion compensation predictor generated as An interlaced scanning moving image encoding apparatus, comprising: a threshold determination offset determiner that calculates and issues the threshold determination offset value from a function for the difference value of. 2. Whether motion compensation is necessary or not is determined by whether a difference value between the current frame image and the locally decoded frame image from the adder is larger than a predetermined threshold value. A motion compensation deciding unit that outputs a locally decoded frame image and issues a coding mode identifying command for intra-frame direct or inter-frame motion compensation prediction, and the motion compensation deciding unit according to the coding mode identifying command from the motion compensation deciding unit. From the current frame image or the locally decoded frame image to an intra-field / inter-field macroblocking processor or a motion compensation predictor, and a coding mode switch for switching between direct and intraframe motion compensation prediction. 2. The interlaced moving image code according to claim 1, wherein any one of the coding modes is adaptively selected. Device. 3. A quantizer is used to directly input the motion compensation result information from the motion compensation predictor as feedforward information from the preceding stage, and perform quantization control in place of or in combination with the feedback information from the succeeding stage. The interlaced scanning moving image encoding device according to claim 1 or 2. 4. A threshold judgment offset value determiner uses motion compensation result information on the macroblock and a plurality of peripheral macroblocks from a motion compensation predictor, and a subtracter judges each threshold on a macroblock basis of a prediction error frame image. 4. The interlaced scanning moving image encoding apparatus according to claim 1, wherein the offset value is calculated and, for example, each arithmetic mean value is used as the threshold determination offset value. 5. When the threshold judgment offset value determiner causes a large deterioration in image quality of the entire current frame image, instead of using the motion compensation result information on the current frame from the motion compensation predictor, efficient encoding can be performed in the past several frames. The threshold judgment offset value is calculated for each frame of the prediction error frame image from the subtracter using the motion compensation result information regarding the preceding frame.
3. The interlaced scanning moving image encoding device as described in 3 or 4.