CN1284119C - Method for providing format of storing images in high compress performance - Google Patents

Abstract

The present invention relates to a method for obtaining an image storage format VCC with high compensation performance. The present invention is characterized in that in the method of the present invention, vertex chain code image recording, vertex chain code image unit search, image scanning, ternary notation coding and vertex chain code image reduction are used according to a boundary calibration automatic machine and a vertex chain code method. The method of the present invention comprises the following concrete steps: three kinds of vertices on a square dot matrix of an image are respectively represented by 1, 2 and 3, the corresponding vertices are orderly recorded along one direction, and the recording of vertex chain codes to the unit image is realized; the vertex chain codes of the unit image are automatically generated by the boundary calibration automatic machine; the coordinates of the starting point of the vertex chain codes are recorded by a ternary number. The image format VCC obtained with the method has the outstanding advantage that the high compensation performance aspect of the image exceeds all the present image storage formats.

Description

A kind of method of image storage format

Technical field

The present invention relates to a kind of method of image storage format, belong to the computer image processing technology field.

Background technology

The volume of image file is big, and this is a very big trouble in the transmission of image file.As on network, image file is big, and the delivery time is just long, makes that the information that transmits on the network is very crowded.A main performance index of image file is, expresses image with the least possible byte, and promptly the volume of image file is little.Particularly, figure shown in Figure 1 is 442k with the BMP form, is 113k with the IFF form, is 7.72k with the GIF form.Therefore on network, transmit this same width of cloth figure, suppose that the BMP form needs 1 second, then needed only 0.256 second and 0.017 second respectively with IFF form and GIF form so.Because on the present network, the transmission consumption of image file is big, it is just extremely considerable to have the economic benefit that the picture format of high compression performance produces.The image transmission speed is fast, has also improved the quality of network service, and the feasible website of the image storage format of high compression performance of adopting has stronger competitive power.The geography information that obtains on the current satellite, the image file with magnanimity, high performance image storage format can solve the bottleneck problem that image is sent to the ground receiving terminal greatly.In addition militarily, the transfer rate that improves image file is most important, the little confidentiality that also can improve in the transport process of image file volume.

Just because the image storage format of high compression performance is all so important from the military affairs economically, and best picture format such as the GIF of compression performance at present writes Fig. 1 form, and the volume of compressed images still be the 7.72k byte, not really ideal.

Summary of the invention

The objective of the invention is provides a kind of method that obtains the image storage format of high compression performance in order further to improve the compression performance of image.

Technical scheme of the present invention is achieved like this, a kind of method of image storage format, and it is characterized in that according to region labeling automaton and vertex-chain-code method concrete step is:

1, carries out image recording by vertex-chain-code, three kinds of summits are arranged on the square dot matrix of image, with 1,2,3 representatives, write down corresponding summit successively respectively, be the record of vertex-chain-code cell picture along a direction;

2, by automatic search method the border of image is searched for automatically, produced the elementary area vertex-chain-code automatically with region labeling automaton with bianry image;

3, by the sweep record of image, define an augmented matrix and come image is scanned, the vertex-chain-code on the starting point on each border of each figure in the image and border is noted;

4, pass through the coordinate of trit record chain coding and chain encoding starting point, use two scale-of-two as the coding unit, four efficient codings are arranged, represent 1 in the chain encoding respectively for wherein three, 2,3 three sign indicating numbers, a remaining efficient coding is as identifier, simultaneously metric coordinate figure numeral is converted into ternary numeral, comes the coordinate of record chain coding starting point;

5, the display image again on computer screen of the image restoring by vertex-chain-code, it is by utilizing the characteristic of chain encoding, determining that according to the line segment of boundary curve maximum value or minimum value the border fill method of the symbol of symbol device is realized.

The present invention is at vertex-chain-code and disclosed region labeling automaton (Gu Guoqing, old can: Region-Labeling-Automata for Image in Square, Triangular and HexagonalLattice, Advances in Computer and information Technology, The Journal of ThreeDimensional Images, Vol.17, No.1 (2003) 161-165.) the basis on, invent a kind of new picture format, be named as the VCC file.Worked out the software of coding and image demonstration at present for bianry image (being black white image).The outstanding advantage of this picture format is that the high compression performance aspect of image has surpassed present all image storage formats.Be higher than existing recorded aspect the high compression performance of image, and the amplitude that improves is also remarkable especially, and image shown in Figure 1 is tested, and has write this width of cloth figure with the various picture formats on the Photoshop, also write same width of cloth figure with the VCC form, what obtain the results are summarized in the table 1.

Picture format	The volume of image (k byte)	Volume after the compression (k byte)
			EPS	3661.00	81.40
TIF	444.00	10.70
			BMP	442.00	8.93
RAW	440.00	7.63
			TGA	440.00	8.30
IFF	113.00	6.76
			PSD	22.80	9.85
PCX	22.70	6.43
			PICT	21.80	8.00
PDF	14.10	11.1
			PNG	11.50	9.23
GIF	7.72	7.72
			VCC	2.66	1.63

The comparison of the various picture format performances of table 1.

Can be clear that from the data of table 1 for black white image, the compression performance of GIF form is best in the existing picture format, and the still huge unexpectedly amplitude of the VCC form of our definition ground has surmounted it.Transmit this width of cloth figure with the VCC form, the time of being changed can shorten to 0.345 of existing form.All image files can compress with general compressed software, so that further improve transfer efficiency.Third column in the table 1 is the volume of various picture format files after compression.Compression back PCX file minimum is 6.43k.But VCC further dwindles, and reaches 1.63k.The ratio of the two is 3.94: 1, near 4 times.We have chosen 20 width of cloth figure wantonly, and the compression performance of VCC is carried out evaluation studies on the statistical significance.For the GIF form, the volume of these 20 files is 250k, and the volume of VCC is 135k then, has only the closely volume of half.After compressing, the volume of GIF form is 249k, and the volume of VCC is 99k, is 3 original/the last 1.Should be noted that all these picture formats and compression all are to implement, and be therefore very important for scientific research under harmless condition, also have more widely and use.

Description of drawings

Fig. 1 is 1 a selected arbitrarily width of cloth bi-level digital image;

Fig. 2 is the state set of region labeling automaton;

Fig. 3 is the state transition diagram of automat.

Embodiment

Concrete steps are:

1, adopt the method for chain encoding to carry out record images, three kinds of summits are arranged on the square dot matrix of image, with 1,2,3 representatives, write down corresponding summit according to this respectively along a direction, be the record of vertex-chain-code, come document image by the border of recording geometry to cell picture.In square dot matrix image, along the border, can recording geometry by mark boundaries pixel summit;

2, with the automatic search method of bianry image the border of image is searched for automatically, for eight neighbours' image, we have defined following region labeling automaton: the state set of automat as shown in Figure 2.Wherein band * number pixel be the current location of automat, the pixel of band arrow is the position of automat previous moment, arrow is represented the direction of automat walking, black picture element is the border.For state A, the state variation rule of automat as shown in Figure 2.Among the figure, the big arrow left side is the state of current time automat, and the big arrow right side is next state of automat constantly, and the pixel of band small circle need check whether be the border, and the numeral in the square frame is the chain encoding that automat provides.Utilize symmetric property, can obtain state B easily, the state transition diagram of state C and state D is encoded with relevant subchain.According to above method, automat will be automatically in accordance with regulations direction along the border walking of figure, provide the chain encoding of graphic limit simultaneously.

3, the sweep record of image, we have proposed the graph scanning recording method and have come document image.

We define an augmented matrix, and matrix column is all composed 0 value than big 1, the first row of image array.Begin image is scanned from first row of augmented matrix, when scanning runs into the pixel of different value, be the border of figure, the record coordinate, the startup region labeling automaton obtains the chain encoding with the recording geometry border.

The starting point on each border of each figure in the image and boundary chain encoded recording are got off the sweep record method of recording geometry chain encoding of our design that Here it is.

4, the three-shift coding has only been used three marks in the chain encoding, utilizes this advantage can make new graphical format have high ratio of compression.We have adopted a kind of new coding method not coordinate and the chain encoding on recording geometry border.Therefore we are the coding unit with two words, four efficient codings can be arranged, and represent 1,2 in the chain encoding respectively for wherein three, 3 three sign indicating numbers, a remaining efficient coding we just as identifier, be designated as "/".Our coding method is: x1/y1/ chain encoding 1/x2/y2/ chain encoding 2 ... .xn/yn/ chain encoding n.For the coordinate of figure, in order to use above coding rule, we adopt the numeral method of 3 systems.

5, the reduction of image shows, according to picture coding display image again on computer screen, comprises the border of writing out figure and fills in the border.We have adopted new border fill method.Polygon in this method and the graphics is filled similar, and different places is the acquisition methods on polygon (being graphic limit) summit.We utilize the characteristic of chain encoding, at first find out the line segment of representing the boundary curve maximum value or minimum value, and when lining by line scan the echo image, when by the border, if the very big or minimum line segment of sign, the symbol device is reversion not, otherwise reversion.So just can simplify to-fill procedure.

Claims (1)

1, a kind of method of image storage format, it is characterized in that according to region labeling automaton and vertex-chain-code method concrete step is:

(1) carries out image recording by vertex-chain-code, three kinds of summits are arranged on the square dot matrix of image, with 1,2,3 representatives, write down corresponding summit successively respectively, be the record of vertex-chain-code cell picture along a direction;

(2) by automatic search method the border of image is searched for automatically, produced the elementary area vertex-chain-code automatically with region labeling automaton with bianry image;

(3) by the sweep record of image, define an augmented matrix and come image is scanned, the vertex-chain-code on the starting point on each border of each figure in the image and border is noted;

(4) coordinate by trit record chain coding and chain encoding starting point, use two scale-of-two as the coding unit, four efficient codings are arranged, represent 1 in the chain encoding respectively for wherein three, 2,3 three sign indicating numbers, a remaining efficient coding is as identifier, simultaneously metric coordinate figure numeral is converted into ternary numeral, comes the coordinate of record chain coding starting point;

(5) display image again on computer screen of the image restoring by vertex-chain-code, it is by utilizing the characteristic of chain encoding, determining that according to the line segment of boundary curve maximum value or minimum value the border fill method of the symbol of symbol device is realized.

Images