INTRODUCTION Data are the means by which information is conveyed.

Same amount of information can be conveyed by different amounts of data. Data compression aims at delivering information efficiently by using least amounts of data. Data compression implies that certain data are redundant [36]. If n1 and n2 denote the number of information-carrying units in two data sets that represent the same information, then Compression Ratio

Relative Data Redundancy

. 2.1 IMAGE DATA REDUNDANCIES 2.1.1. Coding Redundancy Ex: I = [0 0 0 0 1 1 1 1 2 2 3 3 4 5 6 7] Gray Level 0 1 2 3 4 5 6 7 Code1 000 001 010 011 100 101 110 111 Code2 01 00 101 100 1101 1100 1111 1110

Table 2.1 Gray level of coding redundancy Code2 is more efficient to represent the image I in this case because we use the shortest code words (2 digits) to represent the most frequent gray levels (0 and 1 in this case).

Review of Compression Techniques This example illustrates what coding redundancy means. Notice that the length of each codeword in Code2 is related to the relative frequency of each corresponding gray level. It also implies that if each gray level is equally probable, Code1 is most efficient. In other words, there is no coding redundancy. Code2 is an example of variable-length coding. 2.1.2. Interpixel Redundancy Natural image data has one important property. That is the neighboring pixels are highly correlated. This property makes possible further data compression. Ex: I=[0 0 00000 1 1 111...1].

50 50 If 0 and 1 are used to represent each gray level, then 100 digits are required to represent this binary image I. However, we can represent image I as [0 50 1 50] or in binary form as 00000000| 00110010|00000000|00110010. Totally 32 digits are required. This coding is known as run-length coding. 2.1.3. Psycho visual Redundancy Loss of some information may not be sensed visually. This kind of information is referred to as psycho visual redundancy. The simplest example is that only about 20~30 shades of gray can be discerned by human eyes. Therefore, quantizing an image of 256 gray levels into 64 gray levels is usually acceptable for visualization purpose. Notice that, although in general a viewer wont be able to tell the difference, quantization is a lossy compression[39] .Which means the original image can not be restored completely. However, the variable-length coding for removing the coding redundancy and the run-length coding for removing the interpixel redundancy are lossless compressions[34].

Department of Computer Science & Technology, S.K. University, Anantapur 2

Review of Compression Techniques In other words, the decoded data conveys exact information as that the original data conveys. 2.2. FIDELITY CRITERIA There are two classes of criteria for evaluating a lossy compression algorithm. Objective fidelity criteria and subjective fidelity criteria. For objective fidelity criteria, the root-mean-square error and mean-square signal-tonoise ratio are

Often used . Their definitions are:

Subjective fidelity criteria mean measuring image quality by the subjective evaluation of several human observers. 2.3 IMAGE COMPRESSION MODELS [26,29] EMBED PBrush

Figure.2.1. Image Compression Model The source encoder is responsible for reducing or eliminating any coding, interpixel, or psycho visual redundancies in the input image. Normally,

Department of Computer Science & Technology, S.K. University, Anantapur 3

Review of Compression Techniques it consists of three independent operations as shown below:

(a)

(b) Figure 2.2 (a) Source Encoder (b) Source Decoder The first operation is a mapper that transforms the input data into a format designed to reduce interpixel redundancies in the input. This operation is generally reversible and may (ex: run-length coding) or may not (ex: DFT) reduce directly the amount of data required to represent the image. In the latter case, the mapper transforms the input image into a form whose interpixel redundancies are more accessible for compression in later stages. The second operation is a quantizer that reduces the accuracy of the mappers output to reduce the psycho visual redundancies of the input image. This operation is irreversible. Thus, it must be omitted when lossless compression is desired. The third operation is called symbol encoder. It creates usually a variable-length code to represent the quantizer output by assigning the shortest code words to the most frequently occurring output symbol. This operation is again reversible. Not all of the three operations are required in every compression system. In addition, operations can be merged into a single block in some compression system. Foe instance, in lossy predictive coding, map per and squattier can be combined ad represented by a single block. Department of Computer Science & Technology, S.K. University, Anantapur 4

Review of Compression Techniques The source decoder contains only two components: a symbol decoder and an inverse mapper. Recall that there is no operation can undo the quantization operation. The purpose of channel encoder and decoder is to reduce the impact of channel noise by inserting a controlled form of redundancy[37] into the source encoded data. One of the most popular channel encoding techniques is Hamming codes. It is illustrated below. For example, we want to use Hamming codes to encode a 4-digit word b3b2 b1b0 . The 7-bit Hamming (7, 4) code word h1h2 h3h4 h5 h6 h7 associated with the above 4-digit word is generated by the following rules: h1 = b3 b2 b0 h2 = b3 b1 b0

h3 = b3 ; h5 = b2 ; h6 = b1 ; h7 = b0 where denotes the exclusive OR operation. h1 , h2,and h4 are added bits to introduce controlled redundancies. Instead of sending the 4-bit data b3b2 b1b0 over a noisy channel, the 7-bit data h1h2 h3h4 h5 h6 h7 is sent. The advantage of doing so is that if there is any single error occurring during the transmission of the 7-bit data, the error can be detected and recovered. However, if there is any single error occurring during the transmission of the 4-bit data, there is no way to detect and recover the error. The bits h1 , h2 , and h4 are called the even-parity bits for the bit field b3b2 b0 , b3b1b0 , and b2 b1b0 , respectively. The single-bit error occurred in a Hamming (7, 4) code word can be detected by checking the parity word c4 c2 c1 , where

Department of Computer Science & Technology, S.K. University, Anantapur 5

Review of Compression Techniques c1 = h1 h3 h5 h7

c4 = h4 h5 h6 h7 Let i be the decimal representation of the binary number c4 c2 c1 . If i is not equal to zero, we know that there is an error at the ith bit and therefore, the error can be recovered by changing that bit from 0 to 1 or 1 to 0. 2.4 DATA COMPRESSION MODELS During compression, data that is duplicated or that has no value is eliminated or saved in a shorter form, greatly reducing a files size. For example, if large areas of the sky are the same shade of blue, only the value for one pixel needs to be saved along with the locations of the other pixels with the same color. When the image is then edited or displayed, the compression process is reversed.

Figure.2.3 Data Compression and image reconstruction model There are two forms of compressionlossless and lossyand digital cameras use both forms. Lossless Compression Lossless compression technique reconstructs the image so that its quality matches the original source- nothing is lost. Although lossless Department of Computer Science & Technology, S.K. University, Anantapur 6

Review of Compression Techniques compression sounds ideal, it doesnt provide much compression and files remain quite large. For this reason, lossless compression is used mainly where detail is extremely important, as it is when planning to make large prints. Lossless compression is offered by some digital cameras in the form of TIFF and RAW file formats. Lossy Compression Because lossless compression isnt practical in many cases, all popular digital cameras offer a lossy compression (rhymes with "bossy"). This process degrades images to some degree. Images are more degraded if they are compressed more. In many situations, such as posting images on the Web or making small to medium sized prints, the image degradation isn't obvious. However, if you enlarge an image enough, it will show the image degradation. 2.4.1 Comparison between Lossless and Lossy Compression In lossless compression schemes, the reconstructed image, after compression, is numerically identical to the original image. However lossless compression can only a achieve a modest amount of compression. An image reconstructed with lossy compression contains degradation relative to the original. Often this is because the compression scheme completely discards redundant information. However, lossy schemes are capable of achieving much higher compression. Under normal viewing conditions, no visible loss is perceived (visually lossless). 2.5 LOSSLESS COMPRESSION TECHNIQUES 2.5.1. Run-Length Encoding (RLE) Run-length encoding [34] is a very simple method for compression of sequential data (and particularly for binary data) that takes advantage of the fact that consecutive single tokens (gray values) are often identical in many data Department of Computer Science & Technology, S.K. University, Anantapur 7

Review of Compression Techniques sets. Run length encoding inserts a special token each time a chain of more than two equal input tokens are found. This special input advises the decoder to insert the following token n times into his output stream. For example, consider the sequence of 3-bit data.

The encoded RLE image would be: 7321652259 where the first digit in the encoded image is the gray value of the first pixel, the second digit is the number of occurences, etc. The sequence is reduced from 20 3-bit numbers (60 bits) to 10 digits (though the long run of 5 at the end requires a 4-bit number to encode). This means that this sequence might be encoded into 5 3-bit numbers and 5 4-bit numbers, for a total of 35 bits. If the system were limited to three bits, the sequence of 9 examples of level 5 would be split into one sequence of 7 and one of 2: 732165225752 for a total of 10 3-bit numbers, or 30 bits. The compression in RLE occurs when the image exhibits strings of the same value. If the image is noisy, then the image after RLE coding will likely require more bits than the uncompressed image. The common bitmap image format (.BMP) uses run-length encoding. 2.6 LOSSY COMPRESSION TECHNIQUES 2.6.1 Predictive Compression Methods Predictive compressions use image information redundancy (correlation of data) to construct an estimate ~f(i,j) of the gray level value of an image element (i,j) from values of gray levels in the neighborhood of (i,j). In image parts where data are not correlated, the estimate ~f will not match the original value.The differences between estimates and reality, which may be expected to Department of Computer Science & Technology, S.K. University, Anantapur 8

Review of Compression Techniques be relatively small in absolute terms, are coded and transmitted together with prediction model parameters -- the whole set now represents compressed image data.The gray value at the location (i,j) is reconstructed from a computed estimate ~f(i,j) and the stored difference d(i,j) is

Differential Pulse Code Modulation (DPCM) is shown Fig 2

(a) (b) Figure.3.2. Differential Pulse Code Modulation (a) Compression (b) Reconstruction Linear predictor of the third order is sufficient for estimation in a wide variety of images. The estimate ~f can be computed as follows.

Where a1,a2,a3 are image prediction model parameters. These parameters are set to minimize the mean quadratic estimation error e,

and the solution, assuming f is a stationary random process with a zero mean, using a predictor of the third order, is

Department of Computer Science & Technology, S.K. University, Anantapur 9

Review of Compression Techniques where R(m,n) is the autocorrelation function of the random process f.

(a)

(b)

( c)

(d)

Figure. 2.5 Predictive Compression: (a) Reconstructed Image (k=3.8) (b) Difference Image (k=3.8) (c) Reconstructed Image (k=6.2) (d) Difference Image (k=6.2) 2.6.2 Hierarchical and Progressive Compression Techniques A substantial reduction in bit volume can be obtained by merely representing a source as a pyramid. Approaches exist for which the entire pyramid requires data volume equal to that of the full resolution image. Even more significant reduction can be achieved for images with large areas of the same gray level if a quadtree coding scheme is applied.

Department of Computer Science & Technology, S.K. University, Anantapur 10

Review of Compression Techniques

Figure. 2.6 Principle of quad tree image compression original image and corresponding quad tree Nevertheless, there may be an even more important aspect connected with this compression approach - the feasibility of progressive image transmission and the idea of smart compression. Progressive image transmission - transmitting all image data may not be necessary under some circumstances e.g., searching an image database looking for a particular image. This approach is also commonly used to decrease the waiting time needed for the image to start appearing after transmission and is used by World Wide Web image transmissions. In progressive transmission, the images are represented in a pyramid structure, the higher pyramid levels (lower resolution) being transmitted first. The concept of smart compression is based on the sensing properties of human visual sensors. The spatial resolution of the human eye decreases significantly with increasing distance from the optical axis. Therefore, the human eye can only see in high resolution in a very small area close to the point where the eye is focused. Similarly as with image displays, where it does not make sense to display or even transmit an image in higher resolution than that of the display device, it is not necessary to display an image in full resolution in image areas where the user's eyes are not focused. The main difficulty remains in determining the areas of interest in the image on which the user will focus. When considering a smart progressive image transmission, the image should be transmitted in higher resolution in areas of interest first - this improves a subjective rating of transmission speed as sensed by a human user. The areas of interest may be obtained in a feedback control manner from tracking the user's eyes (assuming the communication channel is fast enough). This smart image transmission and compression may be extremely useful if Department of Computer Science & Technology, S.K. University, Anantapur 11

Review of Compression Techniques applied to dynamic image generators in driving or flight simulators, or to high definition television. 2.7 DISCRETE COSINE TRANSFORM (DCT) Uncompressed multimedia (graphics, audio and video) data requires considerable storage capacity and transmission bandwidth. Despite rapid progress in mass-storage density, processor speeds, and digital communication system performance, demand for data storage capacity and data-transmission bandwidth continues to outstrip the capabilities of available technologies. The recent growth of data intensive multimedia-based web applications have not only sustained the need for more efficient ways to encode signals and images but have made compression of such signals central to storage and communication technology. For still image compression, the `Joint Photographic Experts Group' or JPEG[23] standard has been established by ISO (International Standards Organization) and IEC (International Electro-Technical Commission). The performance of these coders generally degrades at low bit-rates mainly because of the underlying block-based Discrete Cosine Transform (DCT)[31] scheme. More recently, the wavelet transform has emerged as a cutting edge technology, within the field of image compression. Wavelet-based coding provides substantial improvements in picture quality at higher compression ratios. Over the past few years, a variety of powerful and sophisticated wavelet-based schemes for image compression, as discussed later, have been developed and implemented. Because of the many advantages, the top contenders in the upcoming JPEG-2000 standard [7,8] are all wavelet-based compression algorithms. The goal of this article is two-fold. First, for readers new to compression, we briefly review some basic concepts on image compression and present a short overview of the DCT-based JPEG standard and the more popular wavelet-based image coding schemes. Second, for more advanced readers, we Department of Computer Science & Technology, S.K. University, Anantapur 12

Review of Compression Techniques mention a few sophisticated, modern, and popular wavelet-based techniques including one we are currently pursuing. The goal of the upcoming JPEG-2000 image compression standard, which is going to be wavelet-based, is briefly presented. For those who are curious, a number of useful references are given. There is also abundance of information about image compression on the Internet. The figures in Table 2.3 show the qualitative transition from simple text to full-motion video data and the disk space, transmission bandwidth, and transmission time needed to store and transmit such uncompressed data. Table 2.3 Multimedia data types and uncompressed storage space,

transmission bandwidth, and transmission time required. The prefix kilodenotes a factor of 1000 rather than 1024.

Table.2.3 Qualitative transition from simple text to full-motion video data and the disk space The examples above clearly illustrate the need for sufficient storage space, large transmission bandwidth, and long transmission time for image, audio, and video data. At the present state of technology, the only solution is to compress multimedia data before its storage and transmission, and decompress it at the receiver for play back. For example, with a compression ratio of 32:1, Department of Computer Science & Technology, S.K. University, Anantapur 13

Review of Compression Techniques the space, bandwidth, and transmission time requirements can be reduced by a factor of 32, with acceptable quality. A common characteristic of most images is that the neighboring pixels are correlated and therefore contain redundant information. The foremost task then is to find less correlated representation of the image. Two fundamental components of compression are redundancy and irrelevancy reduction. Redundancy reduction aims at removing duplication from the signal source (image/video). Irrelevancy reduction omits parts of the signal that will not be noticed by the signal receiver, namely the Human Visual System (HVS). In general, three types of redundancy can be identified: Spatial Redundancy or correlation between neighboring pixel values. Spectral Redundancy or correlation between different color planes or spectral bands. Temporal Redundancy or correlation between adjacent frames in a sequence of images (in video applications). Image compression research aims at reducing the number of bits needed to represent an image by removing the spatial and spectral redundancies as much as possible. Since we will focus only on still image compression, we will not worry about temporal redundancy. 2.7.1 Image Coder A typical lossy image compression system is shown in Fig. 2.7 It consists of three closely connected components namely (a) Source Encoder (b) Quantizer, and (c) Entropy Encoder. Compression is accomplished by applying a linear transform to decorrelate the image data, quantizing the resulting transform coefficients, and entropy coding the quantized values.

Department of Computer Science & Technology, S.K. University, Anantapur 14

Review of Compression Techniques Figure.2.7. A Typical Lossy Signal/Image Encoder 2.7.2 Source Encoder (or Linear Transformer) Over the years, a variety of linear transforms have been developed which include Discrete Fourier Transform (DFT), Discrete Cosine Transform (DCT) , Discrete Wavelet Transform (DWT) and many more, each with its own advantages and disadvantages. 2.7.3 Quantizer A quantizer simply reduces the number of bits needed to store the transformed coefficients by reducing the precision of those values. Since this is a many-to-one mapping, it is a lossy process and is the main source of compression in an encoder. Quantization can be performed on each individual coefficient, which is known as Scalar Quantization (SQ). Quantization can also be performed on a group of coefficients together, and this is known as Vector Quantization (VQ)[30]. Both uniform and non-uniform quantizers can be used depending on the problem at hand. For an analysis on different quantization schemes, see

Table 2.4 Quantization

2.7.4 Entropy Encoder

Department of Computer Science & Technology, S.K. University, Anantapur 15

Review of Compression Techniques An entropy encoder further compresses the quantized values losslessly to give better overall compression. It uses a model to accurately determine the probabilities for each quantized value and produces an appropriate code based on these probabilities so that the resultant output code stream will be smaller than the input stream. The most commonly used entropy encoders are the Huffman encoder and the arithmetic encoder, although for applications requiring fast execution, simple run-length encoding (RLE) has proven very effective. It is important to note that a properly designed quantizer and entropy encoder are absolutely necessary along with optimum signal transformation to get the best possible compression. 2.7.5 JPEG: DCT-Based Image Coding Standard The idea of compressing an image is not new. The discovery of DCT in 1974 is an important achievement for the research community working on image compression. The DCT can be regarded as a discrete-time version of the Fourier-Cosine series. It is a close relative of DFT, a technique for converting a signal into elementary frequency components. Thus DCT can be computed with a Fast Fourier Transform (FFT) like algorithm in O(n log n) operations. Unlike DFT, DCT is real-valued and provides a better approximation of a signal with fewer coefficients. The DCT of a discrete signal x(n), n=0, 1, .. , N-1 is defined as:

where, C(u) = 0.707 for u = 0 and =

otherwise.

An excellent analysis of DCT and related transforms and their applications can be found in [26,29]. In 1992, JPEG [26] established the first international standard for still image compression where the encoders and decoders are DCT-based. The JPEG

Department of Computer Science & Technology, S.K. University, Anantapur 16

Review of Compression Techniques standard specifies three modes namely sequential, progressive, and hierarchical for lossy encoding, and one mode of lossless encoding. The `baseline JPEG coder' [26,31] which is the sequential encoding in its simplest form, will be briefly discussed here. Fig. 2.8 show the key processing steps in such an encoder and decoder for grayscale images. Color image compression [10]can be approximately regarded as compression of multiple grayscale images, which are either compressed entirely one at a time, or are compressed by alternately interleaving 8x8 sample blocks from each in turn. In this article, we focus on grayscale images only.

(a)

(b) Figure. 2.8 (a) JPEG Encoder Block Diagram (b) JPEG Decoder Block Diagram The DCT-based encoder can be thought of as essentially compression of a stream of 8x8 blocks of image samples. Each 8x8 block makes its way through each processing step, and yields output in compressed form into the data stream. Because adjacent image pixels are highly correlated, the `forward' DCT (FDCT) processing step lays the foundation for achieving data Department of Computer Science & Technology, S.K. University, Anantapur 17

Review of Compression Techniques compression by concentrating most of the signal in the lower spatial frequencies. For a typical 8x8 sample block from a typical source image, most of the spatial frequencies have zero or near-zero amplitude and need not be encoded. In principle, the DCT introduces no loss to the source image samples; it merely transforms them to a domain in which they can be more efficiently encoded. After output from the FDCT, each of the 64 DCT coefficients is uniformly quantized in conjunction with a carefully designed 64-element Quantization Table (QT)[31]. At the decoder, the quantized values are multiplied by the corresponding QT elements to recover the original unquantized values. After quantization, all of the quantized coefficients are ordered into the "zig-zag" sequence as shown in Fig. 2.9 This ordering helps to facilitate entropy encoding by placing low-frequency non-zero coefficients before high-frequency coefficients. The DC coefficient, which contains a significant fraction of the total image energy, is differentially encoded.

Figure. 2.9. Zig-Zag sequence

Entropy Coding (EC) achieves additional compression losslessly by encoding the quantized DCT coefficients more compactly based on their statistical characteristics. The JPEG proposal specifies both Huffman coding and arithmetic coding. The baseline sequential codec uses Huffman coding, but codecs with both methods are specified for all modes of operation. Arithmetic coding, though more complex, normally achieves 5-10% better compression than Huffman coding[38]. Department of Computer Science & Technology, S.K. University, Anantapur 18

Review of Compression Techniques

(a)

(b)

( c)

(d)

Figure 2.10 Discrete Cosine Image Compression (a) Reconstructed Image (k=4.2) (b) Difference Image (k=4.2) (c) Reconstructed Image (k=5.6) (d) Difference Image (k=5.6). 2.8 WAVELETS AND IMAGE COMPRESSION Wavelets are functions defined over a finite interval and having an average value of zero. The basic idea of the wavelet transform is to represent any arbitrary function (t) as a superposition of a set of such wavelets or basis functions. These basis functions or baby wavelets are obtained from a single prototype wavelet called the mother wavelet, by dilations or contractions (scaling) and translations (shifts). The Discrete Wavelet Transform of a finite Department of Computer Science & Technology, S.K. University, Anantapur 19

Review of Compression Techniques length signal x(n) having N components, for example, is expressed by an N x N matrix. For a simple and excellent introduction to wavelets, see [4,20]. For a thorough analysis and applications of wavelets and filter banks, see [5,19,24,25]. Despite all the advantages of JPEG compression schemes based on DCT namely simplicity, satisfactory performance, and availability of special purpose hardware for implementation, these are not without their shortcomings. Since the input image needs to be ``blocked,'' correlation across the block boundaries is not eliminated. This results in noticeable and annoying ``blocking artifacts'' particularly at low bit rates as shown in Fig. 2.11 Lapped Orthogonal Transforms (LOT) [27] attempt to solve this problem by using smoothly overlapping blocks. Although blocking effects are reduced in LOT compressed images, increased computational complexity of such algorithms do not justify wide replacement of DCT by LOT.

Figure. 2.11 (a) Original Lena Image, and (b) Reconstructed Lena with DC component only, to show blocking artifacts Over the past several years, the wavelet transform has gained widespread acceptance in signal processing in general, and in image compression research in particular. In many applications wavelet-based schemes (also referred as subband coding) outperform other coding schemes like the one based on DCT. Since there is no need to block the input image and its basis functions have variable length, wavelet coding schemes at higher compression avoid blocking artifacts. Wavelet-based coding is more robust under Department of Computer Science & Technology, S.K. University, Anantapur 20

Review of Compression Techniques transmission and decoding errors, and also facilitates progressive transmission of images. In addition, they are better matched to the HVS characteristics. Because of their inherent multi resolution nature , wavelet coding schemes are especially suitable for applications where scalability and tolerable degradation are important. 2.8.1 Subband Coding The fundamental concept behind Subband Coding (SBC) [21,22] is to split up the frequency band of a signal (image in our case) and then to code each subband using a coder and bit rate accurately matched to the statistics of the band. SBC has been used extensively first in speech coding and later in image coding because of its inherent advantages namely variable bit assignment among the subbands as well as coding error confinement within the subbands.

Figure.2.12 (a) Separable 4-subband Filterbank, and (b) Partition of the Frequency Domain Woods and O'Neil [35] used a separable combination of onedimensional Quadrature Mirror Filterbanks (QMF) to perform a 4-band decomposition by the row-column approach as shown in Fig. 2.12(a). Corresponding division of the frequency spectrum is shown in Fig. 2.12(b). The process can be iterated to obtain higher band decomposition filter trees. At the decoder, the subband signals are decoded, up sampled and passed through a bank of synthesis filters and properly summed up to yield the reconstructed image Department of Computer Science & Technology, S.K. University, Anantapur 21

Review of Compression Techniques 2.8.2 Subband to Wavelet Coding Over the years, there have been many efforts leading to improved and efficient design of filter banks and subband coding techniques. Since 1990, methods very similar and closely related to subband coding have been proposed by various researchers under the name of Wavelet Coding (WC) using filters specifically designed for this purpose. Such filters must meet additional and often conflicting requirements . These include short impulse response of the analysis filters to preserve the localization of image features as well as to have fast computation, short impulse response of the synthesis filters to prevent spreading of artifacts (ringing around edges) resulting from quantization errors, and linear phase of both types of filters since nonlinear phase introduce unpleasant waveform distortions around edges. Orthogonality is another useful requirement since orthogonal filters, in addition to preservation of energy, implement a unitary transform between the input and the subbands. But, as in the case of 1-D, in two-band Finite Impulse Response (FIR) systems linear phase and orthogonality are mutually exclusive, and so orthogonality is sacrificed to achieve linear phase. 2.8.3 Link between Wavelet Transform and Filterbank Construction of orthonormal families of wavelet basis functions can be carried out in continuous time. However, the same can also be derived by starting from discrete-time filters. Daubechies was the first to discover that the discrete-time filters or QMFs can be iterated and under certain regularity conditions will lead to continuous-time wavelets. This is a very practical and extremely useful wavelet decomposition scheme, since FIR discrete-time filters can be used to implement them. It follows that the orthonormal bases in [32] correspond to a subband coding scheme with exact reconstruction property, using the same FIR filters for reconstruction as for decomposition. So, subband coding developed earlier is in fact a form of wavelet coding in disguise. Wavelets did not gain popularity in image coding until Daubechies established this link in late 1980s. Later a systematic way of constructing a family of

Department of Computer Science & Technology, S.K. University, Anantapur 22

Review of Compression Techniques compactly supported biorthogonal wavelets was developed by Cohen, Daubechies, and Fauve (CDF) . Although the design and choice of various filters and the construction of different wavelets from the iteration of such filters are very important, it is beyond the scope of this article. 2.8.4 Wavelet Decomposition There are several ways wavelet transforms can decompose a signal into various subbands. These include uniform decomposition, octave-band decomposition, and adaptive or wavelet-packet decomposition . Out of these, octave-band decomposition is the most widely used. This is a non-uniform band splitting method that decomposes the lower frequency part into narrower bands and the high-pass output at each level is left without any further decomposition. Fig. 2.13(a) shows the various subband images of a 3-level octave-band decomposed Lena using the popular CDF-9/7 biorthogonal wavelet.

Figure.2.13 (a) Three level octave-band decomposition of Lena image, and (b) Spectral decomposition and ordering. Most of the subband and wavelet coding schemes can also be described in terms of the general framework depicted. The main difference from the JPEG[15,16] standard is the use of DWT rather than DCT. Also, the image need not be split into 8 x 8 disjoint blocks. Of course, many enhancements have been

Department of Computer Science & Technology, S.K. University, Anantapur 23

Review of Compression Techniques made to the standard quantization and encoding techniques to take advantage of how the wavelet transforms works on an image and the properties and statistics of transformed coefficients so generated.

(a)

(b)

(c)

(d)

Figure 2.14 Wavelet Image Compression (a) Reconstructed Image (k=4.2) (b) Difference Image (k=4.2) (a) Reconstructed Image (k=5.6) (b) Difference Image (k=5.6)

2.9 BINARY PLANE TECHNIQUES In Binary plane technique[2] Loss Less Compression Technique and lossy technique are proposed.

Department of Computer Science & Technology, S.K. University, Anantapur 24

Review of Compression Techniques The loss less compression technique is based on spatial domain of the image and it is much suitable for the compression of medical images. The main objective of this technique is to take advantage of repeated values in consecutive pixels positions. For a set of repeated consecutive values only one value is retained The Lossy Binary Plane technique, introduces little loss to achieve more compression rate. The Binary Plane Technique is applied for both monochrome and color images. In all of the techniques the compressed file is maintained in two parts. The first part is bit plane and second is data table. The bit plane is collection of 1s and 0s to represent whether a pixel is repeated or not. The second part is data table, which holds only the necessary pixel values. The bit plane and data table are merged into one file. For further compression , another loss less technique Huffman coding [38] is applied and final form of compressed file is generated. 2.9.1 Binary Plane Technique For Color Images The Binary Plane Technique is also applied for color images by separating the three basic color planes 2.9.1.1 Method of Approach for Color Images For color images, the image is first separated into three planes red, green, and blue. For each plane the Binary Plane Technique used for monochrome images is applied. The processing of the pixels of the three planes is done together.

Department of Computer Science & Technology, S.K. University, Anantapur 25

Review of Compression Techniques

Figure.2.15 Format of Intermediate file for Color Images The Bit Plane and Data Tables are maintained separately for each plane. Finally the Bit Planes and Data Tables of each plane are merged in that order to generate intermediate file as shown in the Figure 2.15 In the reconstruction of the image first the intermediate file is generated from the compressed file. The Bit Plane and Data Tables for red, green, and blue planes are extracted from the intermediate file. By checking each bit of Bit Plane for every color plane separately the image is reconstructed. In fourth coming chapter Block Based Binary Plane Techniques are applied for gray scale (Medical and Non-Medical) images for better compression. The algorithm and sample results are explained.

Department of Computer Science & Technology, S.K. University, Anantapur 26