CN111866513A

CN111866513A - Image compression method and image compressor

Info

Publication number: CN111866513A
Application number: CN201910338932.XA
Authority: CN
Inventors: 刘楷; 黄文聪
Original assignee: Realtek Semiconductor Corp
Current assignee: Realtek Semiconductor Corp
Priority date: 2019-04-25
Filing date: 2019-04-25
Publication date: 2020-10-30

Abstract

An image compression method and an image compressor are disclosed. The image compression method is based on the fixed length code to compress the image data to generate the compressed data, and comprises the following steps: judging whether the characteristic value of the image data meets the condition or not; when the characteristic value meets the condition, only the brightness component of the image data is coded to generate the compressed data; when the characteristic value does not meet the condition, encoding the brightness and chroma components of the image data to generate the compressed data; and storing the compressed data.

Description

Image compression method and image compressor

Technical Field

The present disclosure relates to image compression, and more particularly, to adaptive image compression based on fixed-length codes (FLCs).

Background

The Temporal Noise Reduction (TNR) performs low-pass filtering on the target image by using previous image information (previous frame), so as to achieve the effect of suppressing Noise. Temporal Noise suppression can maintain the image detail texture compared to Spatial Noise Reduction (SNR), which typically blurs the image and loses image detail.

In order to store the previous image information, a huge memory is required to record the information, which significantly increases the hardware cost. Therefore, image compression is usually used to reduce the amount of data to be stored, so as to achieve the advantage of reducing hardware cost. In order to save more hardware cost, it is necessary to increase the compression ratio of the previous image information, however, the image compression method with a large compression ratio is prone to cause image distortion, so that the effect of time domain noise suppression is reduced. Therefore, it is a problem to be solved to reduce hardware or memory cost as much as possible while reducing image distortion as much as possible.

Disclosure of Invention

In view of the deficiencies of the prior art, an object of the present invention is to provide an image compression method and an image compressor.

An image compression method for compressing an image data based on a fixed length code to generate a compressed data is disclosed, the method comprising: judging whether a characteristic value of the image data meets a condition; when the characteristic value meets the condition, only the brightness component of the image data is coded to generate the compressed data; when the characteristic value does not meet the condition, encoding the brightness and chroma components of the image data to generate the compressed data; and storing the compressed data.

The present disclosure further discloses an image compressor for compressing an image data based on a fixed length code to generate a compressed data, which includes a memory, a determining circuit and a calculating circuit. The memory is used for storing the image data. The judging circuit is used for judging whether a characteristic value of the image data meets a condition or not. The calculating circuit is coupled to the memory and the judging circuit, and is used for executing the following actions: when the characteristic value meets the condition, only the brightness component of the image data is coded to generate the compressed data; and when the characteristic value does not meet the condition, encoding the brightness and chroma components of the image data to generate the compressed data.

The image compression method and the image compressor can adaptively select a compression (or encoding) mode according to the characteristics of an image. Compared with the traditional technology, the scheme can simultaneously reduce image distortion and hardware cost.

The features, implementations and functions of the present disclosure will be described in detail with reference to the drawings.

Drawings

FIG. 1 is a functional block diagram of an embodiment of an image processing apparatus according to the present disclosure;

FIG. 2 is a functional block diagram of one embodiment of a compressor of the present disclosure;

FIG. 3 is a flowchart illustrating an embodiment of an image compression method according to the present disclosure; and

fig. 4 is a functional block diagram of another embodiment of the compressor of the present disclosure.

Detailed Description

The technical terms in the following description refer to the conventional terms in the technical field, and some terms are explained or defined in the specification, and the explanation of the some terms is based on the explanation or the definition in the specification.

The disclosure includes an image compression method and an image compressor. Since some of the components included in the image compressor of the present disclosure may be known components alone, the following description will omit details of known components without affecting the full disclosure and feasibility of the embodiments of the apparatus. Furthermore, some or all of the processes of the present image compression method may be in software and/or firmware and may be performed by the image compressor or its equivalent, and the following description of the method embodiments will focus on the content of steps rather than hardware without affecting the full disclosure and feasibility of the method embodiments.

Fig. 1 is a functional block diagram of an embodiment of an image processing apparatus 100. The image processing apparatus 100 includes a memory 105, a processing unit 110, a compressor 120, a decompressor 130 and a frame buffer 140. The processing unit 110 performs temporal domain noise suppression on the current frame with reference to the previous frame to generate an output frame. And outputting the image frame, namely the result of the current image frame after the time domain noise suppression. The processing unit 110 may be a circuit or an electronic component with program execution capability, such as a central processing unit, a microprocessor, or a micro-processing unit, that performs time domain noise suppression by executing program codes or program instructions stored in the memory 105. Time domain noise suppression is well known to those skilled in the art and will not be described herein.

The compressor 120 compresses the output frames and stores the compressed output frames in the frame buffer 140 to reduce the amount of data to be stored. Decompressor 130 reads the compressed data from frame buffer 140 and decompresses the compressed data to produce the previous picture frame. The compressor 120 and the decompressor 130 perform FLC-based compression and decompression operations, respectively, with reference to the global feature values. For details of FLC-based compression and decompression operations, reference may be made to taiwan patent publication No. I612800. The global feature values are generated by the processing unit 110 in dependence on the control signals. The control signals, global feature values, and compressor 120 are discussed in more detail in the following embodiments.

The first embodiment is as follows:

fig. 2 is a functional block diagram of an embodiment of the compressor 120. Fig. 3 is a flowchart illustrating an embodiment of an image compression method according to the present disclosure. The compressor 200 includes a determining circuit 210, a calculating circuit 220, a memory 230, and a bitstream editing circuit 240. The memory 230 may be a line buffer (line buffer) for storing data of a plurality of lines of the output frame, and the size of the memory 230 is related to the size of the block (i.e., window) used by the calculating circuit 220 in the compressing operation. For example, if the block size is NxN pixels, the memory 230 stores N lines of data at a time. The output frame is divided into a plurality of non-overlapping blocks in the compression operation, that is, any pixel in the output frame belongs to only one block. When the compression starts, the calculating circuit 220 obtains the image data of one block from the memory 230 (step S310), and then the determining circuit 210 determines whether the global feature value of the output frame is known (step S320). If the processing unit 110 has provided the global feature value (yes in step S320), the determining circuit 210 takes the global feature value as the feature value of the block (step S340). The situation where the processing unit 110 does not provide the global feature value (i.e., no determination in step S320) will be discussed in another embodiment.

After obtaining the feature value of the block, the determining circuit 210 determines whether the feature value of the block meets the condition (step S350). The determining circuit 210 indicates whether the characteristic value satisfies the condition with the mode signal, and the calculating circuit 220 selects the encoding mode according to the mode signal. In step S360, the calculating circuit 220 encodes a luminance (i.e., Y in the YUV color space) component and a chrominance (i.e., U and V in the YUV color space) component of the image data to generate compressed data. In step S370, the calculation circuit 220 encodes only the luminance component of the image data to generate compressed data, and ignores the chroma component of the image data. The bitstream editing circuit 240 adjusts the data ordering of the compressed data (i.e., the YUV channel FLC encoded data or the Y channel FLC encoded data) so that the decompressor 130 can correctly distinguish the encoding mode during decoding.

The characteristic values and conditions include, but are not limited to, the following three scenarios: (1) the characteristic value is a noise level (noise level) of the image data, and the condition is that the noise level is less than a default value; (2) the characteristic value is a color saturation (color saturation) of the image data, and the condition is that the color saturation is smaller than a default value; and (3) the characteristic value is an image property of the image data, and the condition is whether the image property is a single channel image (one channel image) such as an InfraRed (InfraRed) image.

In scenario (1), the noise level may be known by the processing unit 110 from the control signal, and the control signal may be a gain of a sensor (e.g., a sensor of a camera module). Because the sensor gain (sensor gain) is applied to the entire frame, the feature value of each tile in the output frame is equal to the global feature value of the output frame. The high sensor gain represents the scene or environment with insufficient light source corresponding to the current image frame and has higher noise degree; conversely, a low sensor gain represents a scene or environment with sufficient light sources for the current frame, with a lower noise level. When the feature value does not satisfy the condition (no in step S350), that is, when the noise level is not less than the default value (equivalent to the sensor gain being not less than the threshold value, which is correlated with or equal to the default value), since the color noise is easily perceived, the calculation circuit 220 performs YUV channel FLC encoding (step S360). When the feature value meets the condition (yes in step S350), that is, when the noise level is less than the default value (equivalent to the sensor gain being less than the threshold value, wherein the threshold value is related to or equal to the default value), since the color noise is not easily perceived, the computing circuit 220 performs Y-channel FLC coding to reduce the image distortion (step S370).

In scenario (2), the color saturation is obtained by the processing unit 110 calculating the average value of the color channels of the entire output frame, and the feature value of each block in the output frame is equal to the global feature value of the output frame. When the feature value does not satisfy the condition (no in step S350), that is, when the color saturation is not less than the default value, the calculation circuit 220 performs YUV channel FLC encoding (step S360). When the feature value meets the condition (yes judgment in step S350), that is, when the color saturation is smaller than the default value, the calculation circuit 220 performs Y-channel FLC encoding (step S370).

In scenario (3), the image property is known by the processing unit 110 according to a control signal, which may be generated by a sensor module (not shown) indicating that the current frame is acquired through an RGB sensor or a single-channel image sensor (e.g., an infrared sensor). Because the current frame is generated by the same sensor, the feature value of each tile in the output frame is equal to the global feature value of the output frame. When the feature value does not satisfy the condition (no in step S350), that is, the image property is not a single-channel image, the computing circuit 220 performs the YUV channel FLC coding because there is more color information (step S360). When the feature value meets the condition (yes in step S350), that is, when the image property is a single-channel image, the computing circuit 220 performs Y-channel FLC encoding because there is less color information (step S370).

After the calculation circuit 220 completes encoding, the bitstream editing circuit 240 adapts to the data sequence of the compressed data, and then stores the compressed data into the frame buffer 140 (step S380). Next, the calculating circuit 220 determines whether all blocks in the frame have been encoded (step S390). If step S390 is false, the flow returns to step S310; if step S390 is YES, the process ends, which represents that the compression of one frame has been completed.

Example two:

fig. 4 is a functional block diagram of another embodiment of the compressor 120. The compressor 400 includes a determining circuit 410, a calculating circuit 420, a memory 430, and a bitstream editing circuit 440. The memory 430 may be a line buffer for storing data of a plurality of lines of the output frame, and the size of the memory 430 is related to the size of the block (i.e., window) used by the calculation circuit 420 in the compression operation. The apparatus of fig. 4 is also suitable for the process of fig. 3, and steps S310 and S320 have been described in the previous embodiment, and therefore are not described again. In the present embodiment, when the determination in step S320 is negative, the determining circuit 410 calculates the color channel average value of the image data of the block as the feature value of the block (step S330). In other words, the determining circuit 410 calculates the feature value for each block. The determining circuit 410 then determines whether the characteristic value meets a predetermined condition (step S350), and notifies the calculating circuit 420 of the characteristic value by a mode signal. In this embodiment, the compressor 400 determines the appropriate coding mode for each block; in other words, all blocks of the same frame may not be encoded in the same encoding mode. The bitstream editing circuit 440 adjusts the data ordering of the compressed data (i.e., the YUV channel FLC encoded data or the Y channel FLC encoded data) so that the decompressor 130 can correctly distinguish the encoding modes during decoding.

The following describes how the calculation circuit 220 (or the calculation circuit 420) and the bitstream editing circuit 240 (or the bitstream editing circuit 440) perform steps S360 or S370, taking an example of a block size of 4 × 4 pixels and 8 bits per channel. For details of FLC-based coding, refer to taiwan patent publication No. I612800.

Step S360: assuming that data of Y, U, V three channels of two reference points encoded by the YUV channel FLC are (Y1, U1, V1) and (Y2, U2, V2), respectively (2 x3x8 is 48 bits in total, 2 represents two reference points, 3 represents three channels, and 8 represents 8 bits per channel), the bit stream editing circuit 240 (or the bit stream editing circuit 440) puts the smaller value of the Y channel at the front when editing the bit stream (bitstream); in other words, the bitstream editing circuit 240 (or the bitstream editing circuit 440) determines the sizes of Y1 and Y2 and then sorts them. If Y1< Y2 then the bitstream is: Y1Y2U1U2V1V2+ index table (total bit number: 48+16x 3: 96 bits, 16 represents a block of 4x4 pixels, 3 represents an index value of 3 bits per pixel); if Y2< Y1 then the bitstream is: Y2Y1U2U1V2V1+ index table (total bit numbers: 48+48 ═ 96 bits).

Step S370: the computing circuit 220 (or the computing circuit 420) divides a 4X4 pixel block into 4 sub-blocks of 2X2 pixels (sub-blocks a, b, c, and d) for encoding, so that each block has four sets of reference points (each set has 2X8 being 16 bits, 2 table two reference points, 8 representing Y channel): (Ya1, Ya2), (Yb1, Yb2), (Yc1, Yc2), (Yd1, Yd 2). When Ya1> Ya2, the bitstream editing circuit 240 (or bitstream editing circuit 440) ranks Ya1 in front (Yb, Yc, and Yd are the same). The following is an example: assuming Ya1> Ya2, Yb1> Yb2, Yc1> Yc2, and Yd1> Yd2, the bitstream is: ya1Ya2Yb1Yb2Yc1Yc2Yd1Yd2+ index table (total bit number: 64+16x2 ═ 96 bits, 16 represents a tile of 4x4 pixels, and 2 represents an index value of 2 bits per pixel).

As described above, the bitstream editing circuit 240 (or the bitstream editing circuit 440) determines the position of the luminance component of the reference pixel in the bitstream with reference to the mode signal (i.e., according to whether the feature value meets the condition). The two reference points of the block or the sub-block may be two pixels having the largest and the smallest Y values in the block or the sub-block, respectively.

Note that, when the computing circuit 220 (or the computing circuit 420) finds in step S360 that the two reference points have the same Y value (i.e., Y1 — Y2), the computing circuit 220 (or the computing circuit 420) performs Y channel FLC encoding instead (step S370).

The decompressor 130 has different decoding operations in the first embodiment and the second embodiment. In one embodiment, the decompressor 130 determines to perform YUV channel FLC decoding or Y channel FLC decoding with reference to the global feature value. In the second embodiment, the decompressor 130 selects the highest 16 bits from the bitstream (i.e., the compressed data) and compares the first 8 bits (the first value) of the 16 bits with the last 8 bits (the second value). If the first value is smaller than the second value, the block is encoded by a YUV channel FLC; if the first value is greater than or equal to the second value, the block is encoded via the Y channel FLC. The decompressor 130 performs corresponding decoding operations according to different encoding modes.

The scheme can reduce distortion caused by compression by using different coding modes aiming at different image contents. In the present case, no matter YUV channel FLC coding or Y channel FLC coding, the compressed data obtained after coding has the same data size (for example, the block size is 4 × 4 pixels and each channel is 8 bits, and the compressed data is 96 bits), so the two coding modes can share the memory. In other words, the frame buffer 140 can be shared by the two encoding modes without requiring dedicated memory for each encoding mode. Therefore, the present invention can achieve the best performance of image compression or even time domain noise suppression under the same hardware resource limitation (such as memory space and/or data transmission bandwidth).

The present invention is not limited to the two modes, and may also include FLC coding only for U or V channels, or FLC coding only for U and V channels (i.e. ignoring Y channels).

Because the implementation details and variations of the embodiments of the method of the present invention can be understood by those skilled in the art from the disclosure of the embodiments of the apparatus of the present invention, the repeated description is omitted here for the sake of brevity without affecting the disclosed requirements and the implementability of the embodiments of the method. It should be noted that the shapes, sizes, proportions, and sequence of steps in the drawings are merely illustrative and not intended to limit the scope of the present disclosure, which is understood by those skilled in the art. Furthermore, although the foregoing embodiments are described with reference to time domain noise suppression as an example, the disclosure is not limited thereto, and those skilled in the art can apply the present disclosure to other types of image processing techniques as appropriate.

Although the embodiments of the present invention have been described above, these embodiments are not intended to limit the present invention, and those skilled in the art can apply variations to the technical features of the present invention according to the contents of the present invention, which may fall within the scope of the patent protection sought by the present invention.

[ notation ] to show

100 image processing apparatus

105. 230, 430 internal memory

110 processing unit

120. 200, 400 compressor

130 decompressor

140 frame buffer

210. 410 judging circuit

220. 420 calculation circuit

240. 440 bit stream editing circuit

S310 to S390.

Claims

1. An image compression method for compressing an image data based on a fixed length code to generate a compressed data, the method comprising:

judging whether a characteristic value of the image data meets a condition;

when the characteristic value meets the condition, only the brightness component of the image data is coded to generate the compressed data;

when the characteristic value does not meet the condition, encoding the brightness and chroma components of the image data to generate the compressed data; and

The compressed data is stored.

2. The method of claim 1, wherein the characteristic value is a noise level, an image property or a color saturation of the image data, and the condition is that the noise level is less than a first predetermined value, the image property is a single channel image or the color saturation is less than a second predetermined value.

3. An image compressor for compressing an image data based on a fixed length code to generate a compressed data, comprising:

a memory for storing the image data;

a judging circuit for judging whether a characteristic value of the image data meets a condition; and

a calculating circuit, coupled to the memory and the determining circuit, for performing the following operations:

when the characteristic value meets the condition, only the brightness component of the image data is coded to generate the compressed data; and

when the characteristic value does not meet the condition, the brightness and chroma components of the image data are encoded to generate the compressed data.

4. The image compressor as claimed in claim 3, wherein the characteristic value is a noise level of the image data, and the condition is that the noise level is less than a predetermined value.

5. The image compressor as claimed in claim 3, wherein the characteristic value is an image property of the image data, and the condition is that the image property is a single channel image.

6. The image compressor of claim 3, wherein the characteristic value is a color saturation of the image data, and the condition is that the color saturation is less than a predetermined value.

7. The image compressor as claimed in any one of claims 4 to 6, wherein the image data corresponds to one of a plurality of blocks of a frame, and the blocks have respective feature values.

8. The image compressor of claim 6, wherein the compressed data comprises luminance components of a reference pixel, the compressed data being represented in a bitstream, the image compressor further comprising:

a bit stream editing circuit, coupled to the judging circuit and the calculating circuit, for determining the position of the luminance component of the reference pixel in the bit stream according to whether the feature value meets the condition.

9. The image compressor as claimed in claim 3, wherein encoding only the luminance component of the image data generates a first compressed data, and encoding the luminance and chrominance components of the image data generates a second compressed data, the amount of the first compressed data being equal to the amount of the second compressed data.

10. The image compressor of claim 3, wherein the image data corresponds to one of a plurality of blocks of a frame, and the determining circuit further performs the following operations:

Judging whether a global characteristic value of the image frame is known or not;

when the global characteristic value is known, taking the global characteristic value as the characteristic value of the image data; and

when the global characteristic value is not known, calculating a color channel average value of the image data as the characteristic value of the image data.