Abstract: In order to study the growing law of plant, the authors measured the soybean-leaf area by using computer vision technology without destruction. This paper brings forward a non-destructive measurement method based on bilinear mapping which aimed at problems of how to correct image and wipe off texture feature to conduct non-destructive measurement of soybean-leaf area using computer vision technology. The validity of measurement was not influenced by size, figure differentia and white background of laminae. Experimental results show that laminae-image can be corrected, effective area can be extracted and the influence of texture as well as spot can be wiped off commendably. The accuracy can reach 99% using this method to correct the laminae-image. The study results show that the method based on computer vision technology is accurate, simple and convenient in measuring leaf area, which is very important to collect and use plant information in digital agriculture.