CN103067664B - Image compressing and uncompressing method based on optical principle - Google Patents

Abstract

The invention provides an image compressing and uncompressing method based on an optical principle. The method comprises the following steps: extracting a picture of an image to be processed, processing the extracted picture in a grey mode, modulating the picture through a blazed grating, obtaining a composite picture through the fact that amplitude and phase of pixel points of the picture modulated by the blazed grating are respectively added in a one-point-by-one-point mode, conducting Fourier conversion for the composite picture, obtaining diffraction faculae corresponding to the composite picture, filtering the diffraction faculae in a one-by-one mode through a facula filter, and obtaining an uncompressing picture. The method is simple in operation, low in cost, capable of achieving through a common convex lens or through the fact that a computer virtual optical 4f system is used to simulate the Fourier conversion and an inverse transformation process, and capable of being wide used in the field of picture compressing and transmission. Transmission efficiency is improved, the picture obtained through uncompressing is clear, high in recovery degree, small in error with an original picture, and the method is quite large in application space.

Description

A kind of image compression based on optical principle and decompression method

Technical field

The present invention relates to a kind of image compression based on optical principle and decompression method.

Background technology

Along with the fast development of the multimedia technology such as broadband network and digital movie, the acquisition of single image data, transmission, process can not meet the requirement of people to amount of information far away, how the view data of the large information capacities such as film is carried out compressing and transmit and obtained industry and more and more pay close attention to, and this wherein, Image Compression just becomes the most key core technology.The methods such as the compress techniques such as traditional MPEG can not adapt to its demand for development, the Wavelet Image Compression based on JPEG2000 conventional at present can not be satisfactory completely for the compression artefacts of consecutive image.Optical technology is as the appearance of practical technique means undoubtedly for our research opens a new road, but the physical operations of optical technology also exists again many difficulties in force.The price comparison of such as operating equipment is expensive, and the precision requirement of optical instrument is higher, realizes difficulty large, and operation implements complicated difficult to be applied to the problems such as actual.

Summary of the invention

The present invention proposes a kind of image compression based on optical principle and decompression method, its object is to, overcome operation in above-mentioned prior art and implement complicated, the expensive problem of operating equipment.

Technical scheme that the present invention adopts is as follows:

Based on image compression and the decompression method of optical principle, comprise the steps:

Step 1: extract the image in pending image successively, and gray proces is carried out to image, number of image frames is n, and wherein n is integer;

Step 2: adopt the n two field picture that balzed grating, modulation step 1 obtains;

Step 3: being added according to pixels being put pointwise respectively mutually by the amplitude of each pixel of the n two field picture of Grating Modulation and position in step 2, obtaining the composograph that a width comprises amplitude and position phase;

Step 4: front focal plane composograph being placed on convex lens, convex lens are utilized to have the character of two-dimensional Fourier transform, namely light beam passes through convex lens, the image that focal plane is formed into comprises the frequency space information of this Shu Guang, namely obtain Fourier transform results at the back focal plane of convex lens, obtain the diffraction pattern distribution map of frequency spectrum plane;

Convergent lens (i.e. convex lens) is except having imaging character, and the most useful is exactly that it also has the ability of carrying out two-dimension fourier transform; From physical optics, the Fu Lang under the plane light wave vertical illumination of unit amplitude and fraunhofer-diffraction, realize the Fourier transform of diffraction screen transmittance function just; Namely light beam is the frequency space information of this Shu Guang by the image that convex lens collect on focal plane, that is mathematically carries out the result after a Fourier transform to this Shu Guang; So Fourier transform physical significance is here exactly that the spatial distribution of light is converted to frequency distribution (phase space), being image low frequency part in the part near initial point, is image HFS away from initial point part.

The angle that the anglec of rotation of grating makes folded by grating line direction and vertical direction for being rotated counterclockwise grating, and the anglec of rotation determination diffraction pattern of grating is in the angular position of frequency spectrum plane (or diffraction plane);

The flare angle of grating is determined by the angle of inclination of each cutting of grating, and the flare angle determination diffraction pattern of grating is in the radial position of frequency spectrum plane (or diffraction plane);

Flare angle is the angle that grating diffration luminous point departs from optical axis, i.e. the angle of diffraction of its+1 order diffraction hot spot.

Because different images is formed by the Grating Modulation that flare angle is different with the anglec of rotation, thus on frequency plane, non-overlapping copies is understood in the spectrum distribution point position of its+1 order diffraction hot spot, and namely each diffraction pattern includes the image information being different from other hot spots.

Step 5: select hot spot filter to be filtered by diffraction pattern one by one; Wherein, the hot spot number of filter that filtration is selected is consistent with the diffraction pattern quantity of the frequency spectrum plane obtained in step 4, and the position of diffraction pattern filter is corresponding respectively with the position of diffraction pattern, and described hot spot filter is logical light circular hole;

Step 6: diffraction pattern filtration obtained successively is separately through convex lens, and namely carry out inverse Fourier transform, the light distribution obtained from the back focal plane of convex lens is decompressing image.

In described step 2, the concrete steps of balzed grating, modulation image are as follows:

Steps A: i represents the i-th two field picture, i=0;

Step B: the vertical direction at grating line place is decided to be prime direction, grating rotating interval angles is a, and the flare angle of grating is b;

Step C: the position of grating is worth mutually and is multiplied with the gray value according to pixels pointwise of the i-th width image in image sequence to be modulated, picture size remains unchanged after being subject to Grating Modulation, and the diffraction pattern of the image after Grating Modulation determines by the flare angle of grating and the anglec of rotation in the position that frequency spectrum plane occurs;

Step D: every angle a rotating grating in the counterclockwise direction, namely progressively increase the anglec of rotation of grating, flare angle remains and do not become b, i=i+1, if grating rotating angle is less than 360 °, then returns step C; Otherwise, enter step e;

Step e: if i<=n, then reduce the rotation steps angle a of grating, increase the flare angle b of grating, return step B, otherwise, terminate Grating Modulation, obtain in frequency spectrum plane, forming a circle diffraction pattern through the diffraction pattern of the image of Grating Modulation, the image that each diffraction pattern correspondence one is modulated; Wherein, the value of grating rotating interval angles a is 1 ° ~ 180 °, and the span of the blaze angle b of grating is 1 ° ~ 89 °;

The selection principle of flare angle makes the overlap of each circle diffraction pattern little as far as possible, and total diffraction pattern number of turns is many as far as possible and often enclose hot spot and be similar to equidistant arrangement in radial direction, until diffraction region is filled up by the diffraction pattern of all modulated images;

Wherein, the functional form of described balzed grating, cutting is as follows;

$g_1\left(x\right)=\exp [j\&CenterDot;\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}\&CenterDot;\mathrm{tg\&theta;}(\cos \mathrm{\&alpha;}+\cos \mathrm{\&beta;})\&CenterDot;x]$

Wherein, α is beam incident angle, and β is light beam reflection angle, and θ is grating cutting angle of inclination, and λ is optical wavelength, x be incident beam in x direction the horizontal range apart from previous incident luminous point, d is screen periods; Incident beam is vertical or incide grating by non-perpendicular fixed angle, therefore the value of α, β is fixing, and the span of screen periods d is 2-100 pixel, g ₁x the span of () is 0 ~ 1.

The flare angle b of the grating in the step B in described step 2 is initially 11.48 °.

Adopt step 4 described in operating automation to the process of step 6:

Composograph is carried out Fourier transform, obtains the diffraction pattern distribution map of frequency spectrum plane;

Diffraction pattern distribution map is placed in the input face of virtual optics 4f system, frequency spectrum plane arranges a spatial filter, this spatial filter makes the diffraction pattern of diverse location pass through successively selectively, and obtain the hot spot after filtering, its essence is exactly that light splitting is filtered into picture;

Virtual optical system utilizes computerized algorithm simulated optical system to the method for optical beam transformation; Optical processing system is made up of the convex lens of two on same optical axis, and the focal length sum being spaced apart them of two convex lens; Virtual optics 4f system utilizes computer, when light beam scioptics, simulation convex lens, to the change action of light beam, are namely made Fourier transform to the light beam being in first lens front focal plane, are then made inverse Fourier transform to the light beam being in second lens front focal plane;

Utilize digital Fourier transform, computing is done to the light field of diffraction pattern after filtering, namely utilizes transmittance function to be multiplied by the complex amplitude of diffraction pattern, then Fourier transform is done to product, get the light distribution of light field after Fourier transform, be decompressing image;

Wherein, spatial filter is transmittance function, and its expression-form is as follows:

$T(f_x,f_y)=\left\{\begin{array}{c}1,\sqrt{{(f_x-f{\mathrm{cx}}_i)}^2+{(f_y-{\mathrm{fcy}}_i)}^2}\&le;R_i\\ 0,\sqrt{{(f_x-{\mathrm{fcx}}_i)}^2+{(f_y-{\mathrm{fcy}}_i)}^2}>R_i\end{array}\right.$

Wherein, f _x, f _yfor the coordinate of spectrum space, fcx _i, fcy _ibe the centre coordinate of i-th round spot filter, R _iit is the radius of i-th round spot filter circular hole.

The radius span of the hot spot filter in described step 5 is 10 ~ 15 pixels, makes the root-mean-square error of decompressing image and original image be 0 ~ 0.3;

Wherein, the root-mean-square error of image is defined as square root sum square of light intensity difference between decompressing image and all pixels of original image and the subduplicate ratio of original image all picture elements light intensity quadratic sum.

Beneficial effect

The present invention proposes a kind of image compression based on optical principle and decompression method, by extracting the image in pending image, and after gray proces is carried out to the image extracted, balzed grating, is utilized to modulate image, composograph is obtained after utilizing the amplitude of each pixel in the image of balzed grating, modulation and position to be added by pixel respectively mutually, synthetic images obtains diffraction pattern corresponding to composograph after carrying out Fourier transform, utilize hot spot filter to filter one by one diffraction pattern, obtain decompressing image; The method is simple to operate, cost is low, utilize conventional convex lens or adopt computer virtual Optical processing system simulation Fourier transform and inverse transformation process to realize, can be widely used in image compression and transmission field, improve efficiency of transmission, through the clear picture obtained that decompresses, recovery extent is high, less with the error of original image, there is great application space.

Accompanying drawing explanation

Fig. 1 is the flow chart of the inventive method;

Fig. 2 is the parts of images extracted continuously from film;

Fig. 3 is balzed grating, schematic diagram;

Fig. 4 is the composograph of all images after Grating Modulation;

Fig. 5 is the diffraction pattern distribution map of composograph in frequency spectrum plane;

Fig. 6 is hot spot filter schematic;

Fig. 7 is for the decompressing image of Fig. 2;

Fig. 8 is the root-mean-square error curve chart of decompressing image.

Embodiment

Below in conjunction with accompanying drawing, the present invention is described in further details.

As shown in Figure 1, for the present invention is a kind of based on the image compression of optical principle and the flow chart of decompression method, comprise the steps:

Step 1: extract the image in pending image successively, and gray proces is carried out to image, number of image frames is n, and wherein n is integer;

For the image extracted continuously from film, carry out compression and decompression operation to the image in film video, n is 198;

As shown in Figure 2, be 12 two field pictures before extracting according to the order of sequence from film, exemplarily.

Step 2: adopt 198 two field pictures that balzed grating, modulation step 1 is extracted, its specific operation process is as follows:

Steps A: i represents the i-th two field picture, i=0;

Step B: the vertical direction at grating line place is decided to be prime direction, grating rotating interval angles is a, and the flare angle of grating is b;

Step C: the position of grating is worth mutually and is multiplied with the gray value according to pixels pointwise of the i-th width image in image sequence to be modulated, after image is subject to Grating Modulation, size still remains unchanged, and the diffraction pattern of the image after Grating Modulation determines by the flare angle of grating and the anglec of rotation in the position that frequency spectrum plane occurs;

Step D: every angle a rotating grating in the counterclockwise direction, namely progressively increase the anglec of rotation of grating, flare angle b remains unchanged, i=i+1, if grating rotating angle is less than 360 °, then returns step C; Otherwise, enter step e;

Step e: if i<=n, then reduce the rotation steps angle a of grating, increase the flare angle b of grating, return step B, otherwise, terminate Grating Modulation, obtain in frequency spectrum plane, forming a circle diffraction pattern through the diffraction pattern of the image of Grating Modulation, the image that each diffraction pattern correspondence one is modulated; Wherein, the value of grating rotating interval angles a is 1 ° ~ 180 °, and the span of the blaze angle b of grating is 1 ° ~ 89 °;

The functional form of described balzed grating, cutting is as follows;

$g_1\left(x\right)=\exp [j\&CenterDot;\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}\&CenterDot;\mathrm{tg\&theta;}(\cos \mathrm{\&alpha;}+\cos \mathrm{\&beta;})\&CenterDot;x]$

Wherein, being respectively α is beam incident angle, and β is light beam reflection angle, and θ is grating cutting angle of inclination, and λ is optical wavelength, and x is the distance of incident beam in x direction, and d is screen periods; Incident beam is vertical or incide grating by fixed angle, therefore the value of α, β is fixing, and screen periods d is taken as 4 pixels, g ₁x the span of () is 0 ~ 1;

As shown in Figure 3, be balzed grating, schematic diagram, incident light 1 is as reference light, and x represents the horizontal range of incident light 2 and incident light 1, and z is the coordinate figure of incident luminous point at vertical direction;

The angle that the anglec of rotation of grating makes folded by grating line direction and vertical direction for being rotated counterclockwise grating, and the anglec of rotation determination diffraction pattern of grating is in the angular position of frequency spectrum plane (or diffraction plane);

The flare angle of grating is determined by the angle of inclination of each cutting of grating, and the blaze angle determination diffraction pattern of grating is in the radial position of frequency spectrum plane (or diffraction plane);

Flare angle is the angle that grating diffration luminous point departs from optical axis, i.e. the angle of diffraction of its+1 order diffraction hot spot.

Utilize computer to design numeric type balzed grating, according to the reflective function of reflection-type balzed grating, and make by the anglec of rotation selecting the score line angle of inclination (namely determining the flare angle of grating) of balzed grating, and rotating grating to change grating the assigned address being appeared at frequency spectrum plane by this grating diffration hot spot.

For first week and second week rotation modulation, in the process utilizing Grating Modulation input picture, first grating line is decided to be prime direction at vertical direction.Then, be worth mutually the position of grating and be multiplied with the according to pixels pointwise of the gray value of piece image, consequently grating is as a carrier wave, has modulated the gray value of input picture.After image is subject to Grating Modulation, size still remains unchanged, but its diffraction pattern determines by the flare angle of grating and the anglec of rotation in the position that diffraction plane occurs.Then, rotate a grating in the counterclockwise direction every 36 °, namely progressively increase the anglec of rotation of grating, simultaneously blaze angle remains 11.48 °, often change grating rotating angle once after, the input picture making grating new with a width is multiplied.Therefore, the image that balzed grating, can be modulated after rotating 360 degrees is in the counterclockwise direction 10 width.This 10 width image is circle+1 order diffraction hot spot at the diffraction pattern of spectrum domain, and the image that each diffraction pattern correspondence one is modulated.

In second time grating rotating, flare angle remains unchanged after being increased to 16.26 °, and grating is first still that vertical direction is placed, and this grating and the 11 is inputted width image is multiplied.Then progressively rotating grating in the counterclockwise direction from vertical direction, the recruitment of the anglec of rotation changes 18 ° into.Like this, after grating rotating one is enclosed, 20 width images just can by 20 anglecs of rotation the different but Grating Modulation that flare angle is identical.The diffraction pattern of these 20 modulated images will surround a larger concentric circles.

The selection principle of flare angle makes the overlap of each circle diffraction pattern little as far as possible, and total diffraction pattern number of turns is many as far as possible and often enclose hot spot and be similar to equidistant arrangement in radial direction, and by that analogy, to the last a circle diffraction pattern does not exceed observed diffraction region.

Similarly, we continue to increase flare angle and rotating grating, and make the diffraction pattern of the modulation image inputted subsequently to more outer ring expansion, the operation before then repeating, until diffraction region is filled up by the diffraction pattern of all modulated images.According to said method, 198 width images are had modulated.

Step 3: the amplitude of each pixel of the image of all Grating Modulation is according to pixels put pointwise and be added, obtain a width composograph, its amplitude as shown in Figure 4;

Step 4: front focal plane composograph being placed on convex lens, obtains Fourier transform results at its back focal plane, obtains the diffraction pattern distribution map of frequency spectrum plane, as shown in Figure 5;

Because different images is formed by the Grating Modulation that flare angle is different with the anglec of rotation, thus on frequency plane, non-overlapping copies is understood in the spectrum distribution point position of its+1 order diffraction hot spot, and namely each diffraction pattern includes the image information being different from other hot spots.

Diffraction pattern is in orderly the arranging of frequency spectrum plane as can see from Figure 5, because different images is formed by the Grating Modulation that flare angle is different with the anglec of rotation, thus on frequency plane, non-overlapping copies is understood in the spectrum distribution point position of its+1 order diffraction hot spot, and namely each diffraction pattern includes the image information being different from other hot spots.

As can be seen from Diffraction fringe distribution figure we, because each diffraction pattern need occupy certain area, the number of+1 order diffraction hot spot that therefore diffraction surfaces comprises is limited, and the image volume that namely composograph can comprise is limited.The reasonable layout of diffraction pattern will be conducive to us and recover often opening image in filtering, thus realize whole compression and decompression procedure.This depends primarily on the following aspects:

1) spacing of hot spot in same circle;

Determined by grating rotating angle, the radius of circle that in this example, innermost circle hot spot surrounds is 50 pixels, and hot spot number is 10, and therefore spot separation is about 35 pixels;

2) spacing of inner ring and outer ring;

Determined by blaze of grating angle, in this example, the spacing of first lap and the second circle is about 40 pixels from inside to outside;

3) scope of frequency spectrum plane and distribution effective light spot number within it;

Contained by this example intermediate frequency spectrum plane, pixel number is counted out consistent with image pixel, and the effective light spot number be distributed in frequency spectrum plane is determined divided by each hot spot area occupied by frequency spectrum area;

Step 5: select hot spot filter to be filtered by diffraction pattern one by one;

Wherein, the hot spot number of filter that filtration is selected is consistent with the diffraction pattern quantity of the frequency spectrum plane obtained in step 4, and the position of diffraction pattern filter is corresponding respectively with the position of diffraction pattern, and described hot spot filter is logical light circular hole;

Because the position of the position of circular hole each in the filter diffraction pattern corresponding to spectrum domain with size and size match, by a certain circular hole in choice for use filter, certain corresponding for this circular hole diffraction pattern can filter out by we.We know, this diffraction pattern correspond to a modulated two field picture.

The size of filter also governs the number that same frequency spectrum area can hold effective light spot.If the distance between hot spot diminishes, filter size is constant will introduce crosstalk, and the Quality of recovery of image is affected; If spacing is too large, the decrement of image information can be reduced again.

Given this, quality two aspect of least mean-square error to image store capacity and decompressing image of the image after also decompressing after we utilize filtering balances, on the basis of guaranteeing image decompressor quality, namely improve the memory capacity of image information as much as possible.The root-mean-square error value of setting decompressing image and original image is within 0.2, namely the error of decompressing image and original image is no more than 20%, if this value is established too low, then the area requirements of filter is larger, therefore the combination picture capacity that can store is less, if this value establishes get Tai Gao, then the noise of decompressing image is too large, contrast is low, is difficult to identify.In this example, the root-mean-square error of decompressing image and original image can be identified well within 20%, can hold 198 diffraction patterns thus in the frequency spectrum area of 512x512, namely stores or have compressed 198 width images.

In this example, a stacked compression image can store the information of (or compression) 198 images, and the size of superimposed image is 768KB as calculated, and the size of each input picture is also 768KB, therefore, the number ratio of this composograph and input picture, i.e. compression ratio, reaches 198:1.

The filter used in this example as shown in Figure 6, is made up of many independent hot spot filters.Each hot spot design of filter is logical light circular hole, correspond to and filters some diffraction patterns.Logical light circular hole is similar to low pass filter, all can filter out in frequency spectrum plane lower than the optical information of a certain frequency.

The image that the corresponding width of each diffraction pattern is modulated, its main information is distributed near diffraction pattern, and the frequency of information is higher, and distribution is more away from diffraction pattern center.Thus, low pass filter is used just the information below certain frequency of two field picture corresponding for diffraction pattern all can be filtered out.

In the design of filter, on the one hand, consider that the radius of filter is larger, the diffraction pattern information obtained will be more, image will recover more clear, and but then, the information of contiguous diffraction pattern also can enter this filter, cause the information superposition of unrelated images to target image, target image noise is increased.Therefore, in this example, the radius of all hot spot filters is 15 pixels.

Step 6: diffraction pattern filtration obtained successively, separately through convex lens, namely carries out inverse Fourier transform, obtains decompressing image from the back focal plane of convex lens.

As shown in Figure 7, be 12 width consecutive images that the diffraction pattern of Fig. 5 obtains after the filters filter of Fig. 6.This 12 width image corresponds to the result in Fig. 2 after original input picture decompression.

From Fig. 7, we can see that the consecutive image extracted from film obtains decompressing image after compressing, and the clear picture decompressed out, be easy to identify.By each image temporally sequence reconstruct can recover whole film.The root-mean-square error curve of each image recovered as shown in Figure 8, from figure, we can see that the root-mean-square error of all single images all remains on less than 0.2, and whole line smoothing is steady, ensure that the overall Quality of recovery of film, wherein RMS represents root-mean-square error, Frame represents frame, i.e. every two field picture.

Claims (4)

1., based on image compression and the decompression method of optical principle, it is characterized in that, comprise the steps:

Step 1: extract the image in pending image successively, and gray proces is carried out to image, number of image frames is n, and wherein n is integer;

Step 2: adopt the n two field picture that balzed grating, modulation step 1 obtains;

Step 3: being added according to pixels being put pointwise respectively mutually by the amplitude of each pixel of the n two field picture of Grating Modulation and position in step 2, obtaining the composograph that a width comprises amplitude and position phase;

Step 4: front focal plane composograph being placed on convex lens, convex lens are utilized to have the character of two-dimensional Fourier transform, namely light beam passes through convex lens, the image that focal plane is formed into comprises the frequency space information of this Shu Guang, namely obtain Fourier transform results at the back focal plane of convex lens, obtain the diffraction pattern distribution map of frequency spectrum plane;

Step 5: select hot spot filter to be filtered by diffraction pattern one by one; Wherein, the hot spot number of filter that filtration is selected is consistent with the diffraction pattern quantity of the frequency spectrum plane obtained in step 4, and the position of diffraction pattern filter is corresponding respectively with the position of diffraction pattern, and described hot spot filter is logical light circular hole;

Step 6: diffraction pattern filtration obtained successively is separately through convex lens, and namely carry out inverse Fourier transform, the light distribution obtained from the back focal plane of convex lens is decompressing image;

In described step 2, the concrete steps of balzed grating, modulation image are as follows:

Steps A: i represents the i-th two field picture, i=0;

Step B: the vertical direction at grating line place is decided to be prime direction, grating rotating interval angles is a, and the flare angle of grating is b;

Step C: the position of grating is worth mutually and is multiplied with the gray value according to pixels pointwise of the i-th width image in image sequence to be modulated, picture size remains unchanged after being subject to Grating Modulation, and the diffraction pattern of the image after Grating Modulation determines by the flare angle of grating and the anglec of rotation in the position that frequency spectrum plane occurs;

Step D: every angle a rotating grating in the counterclockwise direction, namely progressively increase the anglec of rotation of grating, flare angle remains and do not become b, i=i+1, if grating rotating angle is less than 360 °, then returns step C; Otherwise, enter step e;

Step e: if i<=n, then reduce the rotation steps angle a of grating, increase the flare angle b of grating, return step B, otherwise, terminate Grating Modulation, obtain in frequency spectrum plane, forming a circle diffraction pattern through the diffraction pattern of the image of Grating Modulation, the image that each diffraction pattern correspondence one is modulated; Wherein, the value of grating rotating interval angles a is 1 ° ~ 180 °, and the span of the blaze angle b of grating is 1 ° ~ 89 °;

Wherein, the functional form of balzed grating, cutting is as follows:

$g_1\left(x\right)=\exp [j\&CenterDot;\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}\&CenterDot;\mathrm{tg\&theta;}(\cos \mathrm{\&alpha;}+\cos \mathrm{\&beta;})\&CenterDot;x]$

Wherein, α is beam incident angle, and β is light beam reflection angle, and θ is grating cutting angle of inclination, and λ is optical wavelength, x be incident beam in x direction the horizontal range apart from previous incident luminous point, d is screen periods; Incident beam is vertical or incide grating by non-perpendicular fixed angle, therefore the value of α, β is fixing, and the span of screen periods d is 2-100 pixel, g ₁x the span of () is 0 ~ 1.

2. the image compression based on optical principle according to claim 1 and decompression method, is characterized in that, the flare angle b of the grating in the step B in described step 2 is initially 11.48 °.

3. the image compression based on optical principle according to claim 2 and decompression method, is characterized in that, adopts step 4 described in operating automation to the process of step 6:

Composograph is carried out Fourier transform, obtains the diffraction pattern distribution map of frequency spectrum plane;

Utilize digital Fourier transform, computing is done to the light field of diffraction pattern after filtering, namely utilizes transmittance function to be multiplied by the complex amplitude of diffraction pattern, then Fourier transform is done to product, get the light distribution of light field after Fourier transform, be decompressing image;

Wherein, spatial filter is transmittance function, and its expression-form is as follows:

$T(f_x,f_y)=\left\{\begin{array}{c}1,\sqrt{{(f_x-{\mathrm{fcx}}_i)}^2+{(f_y-{\mathrm{fcy}}_i)}^2}\&le;R_i\\ 0,\sqrt{{(f_x-{\mathrm{fcx}}_i)}^2+{(f_y-{\mathrm{fcy}}_i)}^2}>R_i\end{array}\right.$

Wherein, f _x, f _yfor the coordinate of spectrum space, fcx _i, fcy _ibe the centre coordinate of i-th round spot filter, R _iit is the radius of i-th round spot filter circular hole.

4. the image compression based on optical principle according to claim 3 and decompression method, it is characterized in that, the radius span of the hot spot filter in described step 5 is 10 ~ 15 pixels, makes the root-mean-square error of decompressing image and original image be 0 ~ 0.3;

Wherein, the root-mean-square error of image is defined as square root sum square of light intensity difference between decompressing image and all pixels of original image and the subduplicate ratio of original image all picture elements light intensity quadratic sum.