Classifications

• • H—ELECTRICITY
  • H03—ELECTRONIC CIRCUITRY
  • H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
  • H03M7/00—Conversion of a code where information is represented by a given sequence or number of digits to a code where the same, similar or subset of information is represented by a different sequence or number of digits
  • H03M7/30—Compression; Expansion; Suppression of unnecessary data, e.g. redundancy reduction
  • H03M7/46—Conversion to or from run-length codes, i.e. by representing the number of consecutive digits, or groups of digits, of the same kind by a code word and a digit indicative of that kind
• • H—ELECTRICITY
  • H03—ELECTRONIC CIRCUITRY
  • H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
  • H03M7/00—Conversion of a code where information is represented by a given sequence or number of digits to a code where the same, similar or subset of information is represented by a different sequence or number of digits
  • H03M7/30—Compression; Expansion; Suppression of unnecessary data, e.g. redundancy reduction
  • H03M7/40—Conversion to or from variable length codes, e.g. Shannon-Fano code, Huffman code, Morse code
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
  • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
  • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
  • H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
  • H04N19/186—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being a colour or a chrominance component
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
  • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
  • H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
  • H04N19/61—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding in combination with predictive coding
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
  • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
  • H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
  • H04N19/63—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding using sub-band based transform, e.g. wavelets
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
  • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
  • H04N19/90—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
  • H04N19/93—Run-length coding
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
  • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
  • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
  • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
  • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
  • H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
  • H04N19/13—Adaptive entropy coding, e.g. adaptive variable length coding [AVLC] or context adaptive binary arithmetic coding [CABAC]

Definitions

• the invention relates generally to data and image compression. More specifically, the invention relates to binary encoding for data and image compression, particularly where adaptive quantization and discrete wavelet transforms are utilized.
• encoding In image and/or data compression, through a process known as encoding, a set of values, such as text or numerical data that are obtained or input externally, can be encoded into binary form (Is or 0s).
• One way of encoding is to simply convert each decimal number or code for text (such as ASCII numerical designations) into a fixed number of bits (word size). For instance, the numbers 23, 128, and 100 would be encoded into binary as the sequence: 00010101 1000000 01100100. This raw or pure binary code serves no further compression function, since it merely takes the data and represents it in binary.
• Such encoding is inefficient where the number of zeroes greatly outweigh the non-zero, and especially where such zero data values are consecutive, creating a large "run" of zeroes in binary.
• Several methods have been developed particularly in the field of digital communications to compress data during binary conversion. Among two widely -used such methods of binary encoding for image or data compression are Huffman Coding and Run- Length Encoding.
• the goal behind Huffman Coding is to minimize ⁇ Li P(y ⁇ ), where P(y is the probability of the value yj occurring in data set S that is to be encoded.
• the codewords are chosen in order to make them distinguishable from each other.
• the data value y is encoded by a data structure having two fields: a Range, (that identifies a set containing 2 Rmge values) and a Pointer indicating a specific value inside Range.
• the field Range is coded using Huffman coding, and Pointer is expressed as a binary number having a size of Range bits.
• the values included in a Range of 1 to n values could be:
• Range 1 : values ⁇ -1, 1 ⁇ (the field Pointer needs 1 bit to indicate a specific value).
• Range 2: values ⁇ -3, -2, 2, 3 ⁇ (the field Pointer needs 2 bits).
• Range n: values ⁇ -2 M +1, -2"+2, ..., -2" "1 , 2 n ⁇ , ..., 2"-2, 2"-l ⁇ (the field Poster needs n bits).
• Table 1 The value 0 is coded only by a word 0 (i.e., the Huffman coding of the Range and no other bits).
• the MHC is naturally designed for a table look-up architecture and thus can be more efficient for both encoding and decoding.
• Zero Run Length Coding is a standard technique for encoding a data set containing a large number of consecutive or "runs" of zero values.
• ZRLC consists of encoding only the values different from zero (using Huffman Coding or some other coding) and then interleaving these codewords by a code that specifies the number of zeroes that, according to a manner known both to the coder and to the decoder, divides two consecutive non-zero values.
• the encoded zero run data is structured using two segments: a run length and non-zero value. For instance, instead of coding the data stream:
• This code (where an indicates a run length of zeroes) indicates that 6 zeroes followed by the value 5, then 3 zeroes followed by the value -6, then 0 zeroes followed by the value 78, ...., etc.
• JPEG Joint Photographic Experts Group
• DCT Discrete Cosine Transform
• JPEG Joint Photographic Experts Group
• an image is divided into blocks of pixels such as 8x8, or 16x16 blocks. These blocks are processed independently of each other and thus, the maximum run-length possible is the size of the block (64 or 256).
• the run-length value is 6 bits or 8 bits wide.
• run- length encoding may held fixed in the number of bits comprising the run-length. Where block-based coding is not utilized, such fixed number of bits for coding each "run" (i.e., the number of consecutive zeroes), can become a serious limit, since the longest value for a run depends from the dimension of the whole image.
• a method comprising determining the number of bits utilized to store a word R, the word R accompanied by a flag, the flag and word R comprising a structure; and encoding the length of a run larger than allowable by the number of bits of the word R even if each of said bits were fully utilized, said encoding achieved by stringing together a plurality of such structures.
• a method for determining which of a plurality of encoding schemes applies to each sub-band and channel of data there is a method for determining which of a plurality of encoding schemes applies to each sub-band and channel of data.
• Figures l(a)-l(e) are the essential data structures for modified zero run length encoding according to one embodiment of the invention.
• FIG. 2 illustrates image processing data flow according to one embodiment of the invention.
• Figure 3 illustrates recovery of a compressed and encoded image according to one embodiment of the invention.
• Figure 4 is a diagram of a modified zero run length encoder according to one embodiment of the invention.
• Figure 5 is a block diagram of an adaptive encoder according to one embodiment of the invention.
• FIG. 6 is a block diagram of an image processing apparatus according to an embodiment of the invention.
• Figure 7 is a system diagram of one embodiment of the invention.
• Figures l(a)-l(e) are the essential data structures for modified zero run length encoding according to one embodiment of the invention.
• the size of the run in the ZRLC may assume a value not higher than 63 and thus can be fixed to 6 bit word. If we consider an entire sub-band generated by a 2-D DWT, the upper limit on the size of the run value depends on the size of the whole image. Therefore, one embodiment of the invention provides a flexible structure for encoding the size of the run defined as the number of zeroes between two non-zero values).
• the structure encoding is composed of a Flag F and a word R ofM bits.
• Figure 1(a) shows the basic format of the encoded data.
• one or more structures of the type (F, R) is followed by an MHC (Modified Huffman Code) of any non-zero values in the data set.
• Each ZRLC structure (F, R) has a one-bit F value and bits RI ...RM-
• the first run length is encoded by one structure (F, R).
• the run length terminates when a non-zero value is encountered.
• the exemplary non-zero value is encoded using two bits MH and MHC 2 .
• a second run of zeroes which exceeds 2 M -2 in length is encoded using two (F, R) structures.
• the next non-zero value is encoded using MHC] (only a single bit), after which another run length of zeroes is encoded with one (F, R) structure only.
• a structure of (0, 1) shown in Figure 1(c) denotes that the following non zero value in the path is adjacent to the previous non-zero value, i.e., a run of R -1 zeroes occurs between two consecutive non zero values.
• a run of 4 zeroes would be encoded by the structure (0,5) as shown in Figure 1(d).
• An example of an r longer than 2 M -2 is shown in Figure 1(e).
• the number 86 (1+r) would encode the fact that 85 zeroes are in the run length which is r.
• 86 10 is 0101 1100
• the run length 250 would be output in the order (l,Bl ⁇ ) (0 > Fl6) > and the run length 4091 would be output in the order (l.Cig) (l,Fi6) (0,Fi6).
• any number of structures in the output order (l,Ro) (l,R ⁇ ) (l,R 1-2 ) (0,R ⁇ . i) may be used where IxM bits is the total number of bits required to encode r+1.
• 1-1 structures have a flag of 1
• the Ith structure has a flag of 0.
• One advantage of the modified ZRLC in one embodiment of the invention is that the size of the run may be unlimited and still encodable using a fixed length structure (F,R) of R having only M bits.
• a look-up table be constructed such that each cell is addressed by y (i.e. the data to be encoded) and contains 6 bits.
• the first 2 bits (the field "size") are read and only "size"+l bits of the field "code” are considered and stored as effective information.
• the most significant and underlined bits are the field "range” coded using Huffman Coding (in this example the ranges 0, 1 and 2 are respectively coded by 0, 10 and U), while the least significant bits constitute the field "pointer".
• the LUT coding the MHC for the non zero values in ZRLC in one embodiment of the invention, one can determine an optimal set of such codes based on an analysis of exemplary images. Appendices A-E show such codes arrived by experimentally considering over 50 different images for each sub-band of each channel as a result of these images being subjected to the image compression scheme used in the DWT PATENT.
• the DWT filters used for transforming the images utilized in the analysis were the 9-7 biorthogonal splines filters, the same DWT utilized in the DWT Patent.
• Figure 2 illustrates image processing data flow according to one embodiment of the invention. It is desirable in digital applications such as still or motion imaging that an original image such as one captured by a digital camera be compressed in size as much as possible while maintaining a certain level of quality prior to its being transferred for decompression and/or display.
• the primary image compression technique disclosed in the DWT Patent has been specifically developed to adaptively utilize the response of the human visual system to color and light to maintain image quality. After this compression, according to one embodiment of the invention, an adaptive binary encoding is applied to further compress the data.
• a raw image that is captured by a digital camera or other similar device will typically be represented in a Bayer pattern.
• the sensor array 200 is a set of pixel locations or that contain at each location, an intensity value of the light incident upon the sensors from the environment/scene being imaged.
• each pixel location of an image such as raw image 200 will have an association with a color plane — Red(R), Green(G), or Blue(B). Since the Bayer pattern has two associated values for every R and B, the Green color plane may be considered as two planes Gl and G2.
• the correlation between color planes R, Gl, G2 and B are exploited such that four "channels" are generated and separately compressed.
• the Gl and G2 associated pixels are passed directly to compression (blocks 212 and 216) and comprise two of the channels.
• the R and B pixels are treated less directly.
• the R pixel value is subtracted by its west neighboring Gl pixel value (block 205).
• This difference (R-Gl) channel is passed to compression (block 210).
• each B associated pixel is subtracted from its east neighboring G2 associated pixel (block 206).
• This difference (B-G2) channel is then passed to compression (block 216).
• Each of the four channels, R-Gl, Gl, G2, and B-G2 are passed to compression blocks 210, 212, 214, and 216, respectively.
• the first process is a 2- Dimensional Discrete Wavelet Transform (2-D DWT).
• the 2-D DWT generates "sub- bands" of the image as shown and described in the DWT Patent.
• a second process known as quantization is also performed.
• Quantization is the procedure of mapping a set of n possible values to a set of m possible, where m ⁇ n. By quantizing, the total number of possible data values for the DWT image data set is reduced.
• Adaptive blocks 220, 222, 224 and 226 encoding arranges (packs) the sub-band data from compression blocks 210, 212, 214 and 216, respectively, so that it has an efficient binary representation.
• Adaptive encoding blocks 220, 222, 224 and 226, according to one embodiment of the invention employ an optimal method of encoding to each sub-band and channel combination generated by the compression process.
• Modified ZRLC as defined in embodiments of the invention, is utilized along with the well-known pure Modified Huffman Coding (MHC).
• modified ZRLC uses a fixed number of bits M for R in the structure (F,R), the number M can be varied according to the sub-band and channel being encoded. For instance, in a sub-band with higher entropy to encoding, a smaller value of M may be chosen.
• Table 3 shows optimal encoding strategies which have been empirically arrived for a given sub-band and channel in the case of a DWT performed on Bayer pattern (raw) images using the 9-7 biorthogonal spline DWT filters and quantizing the resulting DWT coefficients by means of the perceptually lossless thresholding as described in the DWT Patent.
• MHC refers to the well-known Modified Huffman Coding
• ZRLC M refers to modified ZRLC as shown in Figures l(a)-l(e) and Figure 4, where M is the number of bits in the structures encoding the run length.
• ZRLCM encodes non-zero values using an MHC.
• the sub-bands LH K , HL , HH K and LL K refer to the various sub-bands created at each level or iteration K of the DWT (see DWT Patent).
• each sub-band and channel is encoded according to its characteristics.
• Figure 2 shows the example of the image compression technique used as being that presented in DWT Patent, any image compression technique which generates data in stages or sub-regions that have differing characteristics can utilize an adaptive encoding such that encoding is optimized to each.
• the compressed and encoded image data 240 may then be stored onto some medium, transmitted from one system to another or distributed over a communication pathway such as a network. Further, the compressed and encoded image data need not be collected and transferred as a single frame, but can be streamed, encoded value by encoded value, out to its destination.
• the compression ratio which the size of the original image divided by the size of the compressed image, will vary.
• This embodiment of the invention provides for an adaptive encoding process that can serve to advantageously optimize the compression ratio in each channel and sub-band generated by the compression process mentioned in the DWT Patent.
• FIG. 1(a)- 1(e) and Figures 4 and 5 One important aspect of the invention which is described in Figures 1(a)- 1(e) and Figures 4 and 5 is a modified zero run length encoding scheme.
• the modified ZRLC may be utilized in any image or data compression process, the use of the compression process from the DWT Patent is merely exemplary.
• an adaptive encoding procedure provided in another embodiment of the invention that is shown responsive to sub-band properties and channel properties may also be modified to be responsive to other types of stage or sub-region generating image compression.
• Figure 3 illustrates recovery of a compressed and encoded image according to one embodiment of the invention.
• the decoding blocks, inverse quantization blocks and inverse DWT blocks comprise a process which attempts to recover the run image 200 from the compressed and encoded image data 240 (see Figure 2).
• the compressed and encoded image data 240 may be efficiently stored channel by channel and sub-band by sub-band.
• the compressed and encoded channels (R-Gl), Gl, G2 and (B-G2) may be separately decoded and decompressed.
• the data belong to each channel and sub-band is decoded (blocks 310, 312, 314 and 316).
• the adaptive encoding (blocks 220, 222, 224 and 226) shown in Figure 2 necessitates that a decoding appropriate to the type of encoding applied for data of given sub-band and channel, be utilized.
• Each channel and sub-band of data may have been encoded using techniques different from those of other sub-bands, and channels, thus, will need to be decoded taking any differences in encoding technique into account.
• Each channel of decoded data is then decompressed (blocks 320, 322, 324 and 326).
• the decompression consists of two procedures — dequantizing the decoded data and then performing an inverse DWT (IDWT). The decompression stage is more fully discussed in the DWT Patent.
• IDWT inverse DWT
• the adaptive decoding can be implemented as hardware, software or from one or a combination thereof and can be separate physically from the apparatus performing the function of the encoding compression process.
• the basic data flow for image processing that consists of compression and recovery may also include an intermediate transfer from the compression block to the desired system or process which has the desired recovery capability needed.
• Figure 4 is a diagram of a modified zero run length encoder according to one embodiment of the invention.
• a modified zero run length encoder where the number of bits M for the word R is 3 may be implemented by a counter 410 and logic network as shown in Figure 4.
• Counter 410 is utilized to determine the total length of the run in binary.
• Counter 410 which is shown to 12 bits in the example may have any number of bits that is desired in the system.
• Each consecutive "zero" encountered in the data stream will increment the counter by 1 , and upon encountering a non zero value, the counter 410 would be reset (to 000000000001), awaiting the next run of zeroes.
• the counter 410 is shown in Figure 4 in a state where a run length (r) of 000101000010 or 322 zeroes is being encoded.
• the counter 410 holds the value 1+r or 000101000011.
• This run length needs to be represented by data structures (Fj,R ; ) and since M, the length of R i? equals 3, the value in counter 410 is considered 3 -bit segments of which there are 4, RQ, R I ,R and R 3 . Additionally, it may be the case that not all 4 available data structures need to be output.
• each field R, in the data structures (1,R,) is used to compose a single run until a structure (0, R,) is encountered.
• a flag of zero for a structure indicates that all bits of the run size have been encountered.
• OR gate 422 a pair of 4-input OR gates 422 and 424 are provided that have as input the OR gate result of the successive structure.
• one input to OR gate 422 is the output of OR gate 424
• OR gate 424 one input is the output of a three-input OR gate 426.
• the other three inputs of OR gate 422 are the bits of Ri while the other three inputs of OR gate 424 are the bits of R 2 .
• the three bits of R 3 form the input of OR gate 426.
• This cascading OR gate structure implements the ability to output and flag correctly as many structures as are needed to represent the run length.
• the structure (F remedyR,) is output if the state of enable signal En, is 1.
• a two-input AND gates govern each enable signal with the exception of Erio.
• EnO is directly controlled by the signal "STOP & RESET” 440 (i.e., is set to 1 when the a non zero value is encountered and therefore the signal "STOP & RESET” 440 is forced).
• Enj is the output of AND gate 432, while En 2 is the output of AND gate 434 and En 3 is the output of AND gate 436.
• the AND gates have as one input the signal M3 (which is Enrj) and have as the other input the flag F belonging to the successive structure. When that flag F is zero, then the end of the run length number has been encountered.
• the flags Fj are arrived at by considering the output of the OR gates.
• the flag F 3 must always equal zero since it by necessity forms the most significant part of any run length value for the given example of the 12-bit counter.
• the run length is less than or equal to 2 M -2, then according to one embodiment of the invention, only (F 0 ,RQ) would need to be output. In this case, the flag F 0 would need to be 0, according to the data structure definition of Figure 1(d).
• a run length less than or equal to 2 M -2 would indicate that the counter will not have filled the bits in the R h R 2 , and R 3 portions of counter 410.
• OR gate 426, 424 and 422 would all show 0 at their output.
• F 0 would be set to 0, as it should.
• Eni, En 2 and En 3 would also be zero, and thus the structures (F ⁇ ,R ⁇ ), (F 2 ,R 2 ) and (F 3 ,R 3 ) would be disabled and not output.
• the ZRLC encoder may be modified depending upon the choice of M and desired total range of the counter.
• FIG. 5 is a block diagram of an adaptive encoder according to one embodiment of the invention.
• the LL k sub-band, according to the DWT Patent, is transformed in order to generate the other three sub-bands of the level k+1.
• the sub- bands LH k , HL k and HH k are then quantized and coded.
• the final level LL sub-band is quantized and coded.
• the other LLk sub-band do not need quantizing and encoding since they are represented by virtue of being further transformed into sub-bands of the next level.
• a signal(s) Encoding Strategy will indicate according to some predefined matching (such as that shown in the Table 3 above) whether the incoming sub- band is to be encoded using the modified ZRLC presented according to various embodiments of the invention, or encoded using pure MHC.
• the Encoding Strategy is passed to a Decision Block 520 which passes the flow control to one of two logical paths. If the Encoding_Strategy indicates pure MHC, the Decision Block 520 will pass flow control to LUTs (Look-Up Tables) for MHC 530.
• one of the LUTs 530 will be selected and the sub-band data will be encoded according to the MHC.
• the LUTs 530 are a series of tables that contain the mapping or Huffman code for a given input value.
• the Huffman code value obtained from LUTs 530 for the given input sub-band data will become the encoded data 570.
• the modified ZRLCM where M is the number of bits for R in the structure (F, R) which encode the run length, is performed by a comparison of the sub-band data value "y" from the DWT block with zero. If this test is true, a counter of Run Length Encoder 560 increases the value r the run length. When a non-zero y (sub-band value) is encountered, this value is used as an address for the LUTs 540 for ZRLC which contain MHC mappings for that non-zero value.
• Run Length Encoder 560 Prior to the MHC for the non-zero value discovered from LUTs 540 being output to form compressed/encoded data 570, the counter in Run Length Encoder 560 is stopped and the structures (F, R) of M+l bits are output to become encoded run length for compressed/encoded data 570. The counter in Run Length Encoder 560 is then reset to a value 1, contemporaneous with which the MHC of the encountered non-zero value from LUTs 540 is output to compressed/encoded data 570 such that the MHC follows the structures (F, R) output by Run Length Encoder 560. Run Length Encoder 560 is governed also by three other signals: a Reset, an M2 and M3 signals.
• the signal Reset indicates the data of the start of a new sub-band.
• M3 When M3 is asserted, a logic network consistent with an M of 3, such as that shown in Figure 4, will be enabled to operate with the run counter.
• an encoder 560 may be equipped with a plurality of logic networks, each appropriate for a different value of M while all may still use the same counter.
• the coding of the sub-band LL FINAL follows a different data flow. To achieve higher compression simple predictive coding involving quantization and error encoding
• the error value between neighboring or successive values is encoded by an entropy coding scheme.
• These values are coded using pure MHC and thus, use LUTs for
• MHC 530 The complete set of modified ZRLC are MHC encoded data represents a compressed version of the original raw image.
• Patent may be considered and encoded separately by the logic of Figure 5, accessing the proper LUTs 530 or 540 which contain MHC values like that shown in Appendices A-E.
• Figure 6 is a block diagram of an image processing apparatus according to an embodiment of the invention.
• FIG. 6 is a block diagram of internal image processing components of an imaging device incorporating at least one embodiment of the invention including an adaptive encoder.
• a sensor 600 generates pixel components which are color/intensity values from some scene/environment.
• the n-bit pixel values generated by sensor 600 are sent to a capture interface 610.
• Sensor 600 in the context relating to the invention will typically sense one of either R, G, or B components from one "sense" of an area or location.
• the intensity value of each pixel is associated with only one of three (or four if Gl and G2 are considered separately) color planes and may form together a Bayer pattern raw image.
• Capture interface 610 resolves the image generated by the sensor and assigns intensity values to the individual pixels. The set of all such pixels for the entire image is in a Bayer pattern in accordance with typical industry implementation of digital camera sensors.
• a RAM 616 consists of the row and column indices of the dead pixels, which are supplied by the sensor. This RAM 616 helps to identify the location of dead pixels in relation to the captured image.
• a primary compressor 628 receives companded sensor image data and performs image compression such as the DWT based compression discussed in the DWT Patent.
• a RAM 629 can be used to store DWT coefficients and/or quantization thresholds for each channel/sub-band as desired in executing the compression techniques described in the DWT Patent.
• Primary compressor 628 can be designed to provide compressed channel by channel and sub-band by sub-band outputs to adaptive encoder 630.
• Adaptive encoder 630 may be designed similar to the design presented in Figures 4 and 5. According to one embodiment of the invention, adaptive encoder 630 can be equipped to carry out a variety of binary encoding schemes, such as pure MHC and the modified ZRLC presented in other embodiments of the invention.
• a RAM 631 may be configured to store the MHC LUTs and the ZRLC LUTs utilized by the adaptive encoder 630.
• Adaptive encoder 630 provides the encoded and compressed data to storage arrays 640.
• the storage arrays 640 may be designed smaller in size than those of typical non-adaptive encoders that uniformly apply only one of the many possible binary encoding techniques to image data.
• Adaptive encoding is particularly advantageous in the encoding of image data, whether compressed or not, since each image (or image compression by-product such as sub-band) may have unique characteristics that would favor one encoding scheme over another.
• Each of the RAM tables 616, 626, 629 and 631 can directly communicate with a bus 660 so that their data can be loaded and then later, if desired, modified. Further, those RAM tables and other RAM tables may be used to store intermediate result data as needed. When the data in storage arrays 640 is ready to be transferred external to the imaging apparatus of Figure 6 it may be placed upon bus 660 for transfer. Bus 660 also facilitates the update of RAM tables 616, 626, 629 and 631 as desired.
• Figure 7 is a system diagram of one embodiment of the invention.
• a computer system 710 which may be any general or special purpose computing or data processing machine such as a PC (personal computer), coupled to a camera 730.
• Camera 730 may be a digital camera, digital video camera, or any image capture device or imaging system, or combination thereof and is utilized to capture an image of a scene 740.
• captured images are processed by an image processing circuit 732 so that they can be efficiently stored in an image memory unit 734, which may be a RAM or other storage device such as a fixed disk.
• the image contained within image memory unit 734 that is destined for computer system 710 can be according to one embodiment of the invention, stored directly as a compressed and adaptively encoded image. In most digital cameras that can perform still imaging, images are stored first and downloaded later.
• the invention in its various embodiments, particularly in providing a compressed image that is directly converted from the captured 8-bit Bayer pattern, reduces the computation requirements of the camera 730 and the associated costs, allowing for a more inexpensive camera.
• the image processing circuit 732 carries out the compression and adaptive encoding in accordance with one or more embodiments of the invention.
• a compressed and encoded image When downloaded to computer system 710, it may be decoded and then rendered to some output device such as a printer (not shown) or to a monitor device 720.
• Image decoding may be achieved using a processor 712 such as the Pentium (a product of Intel Corporation) Processor and a memory 711, such as RAM, which is used to store/load instruction addresses and result data.
• Computer system 710 may provide to camera 730 the LUTs needed for modified ZRLC and MHC shown in Figures 5 and 6.
• the adaptive encoding and modified ZRLC techniques described above may be achieved in a software application running on computer system 710 rather than directly in camera 730.
• the image processing circuit may advantageously store only the compressed image.
• the application(s) used to perform the encoding and/or decoding after download from camera 730 may be from an executable compiled from source code written in a language such as C++.
• the instructions of that executable file, which correspond with instructions necessary to scale the image may be stored to a disk 718 or memory 711. Further, such application software may be distributed on a network or a computer-readable medium for use with other systems.
• Image processing circuit 732 consists of ICs and other components which execute, among other functions, the adaptive encoding of a compressed image.
• the image memory unit 734 will store the compressed and encoded data. Once all pixels are processed and stored or transferred to the computer system 710 for rendering the camera 730 is free to capture the next image.
• the encoded image data in the image memory unit are transferred from image memory unit 734 to the I/O port 717.
• I/O port 717 uses the bus-bridge hierarchy shown (I/O bus 715 to bridge 714 to system bus 713) to temporarily store the data into memory 711 or, optionally, disk 718.
• Computer system 710 has a system bus 713 which facilitates information transfer to/from the processor 712 and memory 711 and a bridge 714 which couples to an I/O bus 715.
• I/O bus 715 connects various I/O devices such as a display adapter 716, disk 718 and an I/O port 717, such as a serial port.
• I/O devices, buses and bridges can be utilized with the invention and the combination shown is merely illustrative of one such possible combination.
• the modified ZRLC encoding and adaptive encoding may be utilized to perform the compression of ordinary non-image data such as the compression of text files as well.
• data may be stored in a disk, memory 711 or other storage mechanism and can be compressed by virtue of being encoded adaptively.
• it may be possible to detect which portions of the data stream may be encoded using a modified ZRLC and which may be encoded using MHC.
• a modified ZRLC scheme that is capable of representing an arbitrarily large run-length value would be particularly efficient.
• MHC may be more suitable.
• Appendix A Huffman tables for the R channel.
• Appendix B Huffman tables for the Gl and G2 channel.
• Appendix C Huffman tables for the B channel.
• Appendix D Huffman tables for the [R-Gl] channel.
• Appendix E Huffman tables for t e [B-G2] channe.

Landscapes

• Engineering & Computer Science (AREA)
• Multimedia (AREA)
• Signal Processing (AREA)
• Theoretical Computer Science (AREA)
• Compression Or Coding Systems Of Tv Signals (AREA)
• Compression, Expansion, Code Conversion, And Decoders (AREA)

Priority Applications (5)

Applications Claiming Priority (2)

Publications (1)

Family

ID=22327843

Family Applications (1)

Country Status (9)

Cited By (1)

Families Citing this family (65)

Citations (2)

Family Cites Families (9)

• 1998
  • 1998-07-02 US US09/109,475 patent/US6124811A/en not_active Expired - Lifetime
• 1999
  • 1999-07-02 KR KR10-2000-7015124A patent/KR100491333B1/ko not_active IP Right Cessation
  • 1999-07-02 DE DE69942028T patent/DE69942028D1/de not_active Expired - Lifetime
  • 1999-07-02 JP JP2000558612A patent/JP4343440B2/ja not_active Expired - Fee Related
  • 1999-07-02 WO PCT/US1999/015221 patent/WO2000002319A1/en not_active Application Discontinuation
  • 1999-07-02 AU AU48611/99A patent/AU4861199A/en not_active Abandoned
  • 1999-07-02 EP EP99932268A patent/EP1095459B1/en not_active Expired - Lifetime
  • 1999-08-03 TW TW088111283A patent/TW453049B/zh not_active IP Right Cessation
• 2000
  • 2000-12-21 GB GBGB0031324.7A patent/GB0031324D0/en not_active Ceased

Patent Citations (2)

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events