The authors examined several problems on developing a character reading system by using a personal computer connected with a facsimile, and a FAX data analyzing method and a developed program are explained in this paper.
In the system of character recognition by using a facsimile and a personal computer which can be used by farmers themselves, there are essential problems which occur fromthe restriction of capability of facsimiles and personal computers.
But by using and analyzing a MH file directly, and by recognizing a character of limited area, the ploblem of memory and speed of data process by a personal computer can be solved. And also, the processing speed can be improved very much by limiting the kinds of characters to hand written numerals and check marks and by developing a special method for recognizing characters.
Besides, by reading+marks on the corner of the data sheet, the right positions for characters can be calculated and corrected.
A FAX data analyzing program developed in this paper can transfer the one dimensionally compressed data which are read from FAX into the real image data, then the essential environment for developing the character recognition system can be achieved.
A flow chart of the FAX data analysis and developed programs are shown in Fig.5, Fig.6 and List respectively.

View full abstract

The authors have developed a character reading system by using a facsimile, and a method of character recognition which is applied in this system and some executed results are explained in this paper, including a comparison analysis of a general recognition method which is popular as the Patttern Matching Method.
The applied method in this paper is called Vector Method which is characterized by making characterization of the figures of handwritten numerals, and after processing directions and lengths of a character and making pattern classification, the numeric character can be recognized.
By using the Vector Method one can recognize numeric characters not only in a short time but also with a small computer memory. A data-sheet with 115 handwritten data can be processed within 2 minutes and the validity of character recognition is about 90% at pesent.
Therefore it can be concluded that this method is appropriate to the character recognition system for personal computers.
A flow chart of the character recognition procedure is shown in Fig.1 and by way of example, image processing results are shown from Fig.2 to Fig.5, Fig.13 and Fig.17.

View full abstract

We tried to extract response patterns of genotypes to environments by the decomposition of the sum of squares of genotype×environment interaction. This is done by using the fuzzy k-means method to get clusters of interaction and by optimizing the number of clusters and the degree of fuzziness from FGC.
When we applied it to soybean data cultivated in the Chusin Agri. Exp. Sta., we could get two fuzzy patterns for the stem length, three for the 100 seeds weight and six for yield.

View full abstract

I have developed a competitive-producing district analysis system, utilizing markets information data referred to a vegetable and fruits markets information database “NAPASS” (Nation-wide Agricultural Products Price Analysis Support System). This system is a markets' trend analysis system developed by personal computers. This system enab-les the shipment analysis of each item of each producing district on wholesale markets of the whole country, and enables also the distributive load analysis of each market on the producing districts throughout the country.
By these analysises, one can clarify the realities of shipment in vagetable producing districts, and also identify the districts competing to them to clarify the realities of shipment and distributive load markets there. The information obtained from these analysises will be useful in working out marketing strategies of vegetable producing districts.

View full abstract