飞机零位检测系统中的座舱仪表读数识别法

介绍了飞机零位检测系统中的座舱仪表读数自动识别系统。识别系统预先在实验室条件下将各分立仪表图象二值化细化处理成标准的预模式，以此为参照输入被测仪表图象，使之与标准模式有接近的几何位置，然后采用二进小波变换方法在不同角度的扇形区域选择不同的变换参数提取仪表指针及刻度信息，抑制噪声与仪表表面玻璃的反光干扰。再建立图象的同心几何结构特征空间，从而求出被测仪表图象的中心点和相对于标准模式的旋转角度，最后判读指针角度，根据此角度与读数的对应关系自动识别出仪表读数。本文建立的系统对外场环境下摄入的带噪声和反光干扰的座舱仪表图象进行自动识别取得了满意的结果，识别精度高于人工识别精度，基本实现了在线测试。

Recognition of Readings on Cockpit Instrument Dial for Ground Test System

The recognition system of cockpit instrument dial readings for ground test system isintroduced. First,evety type of instrument dial images is processed to form standard premodelin the good laboratory conditions. It is used as a frame of reference to grab the tested dialimage in the similar position of the premodel. In order to decrease the influence of noise and reflection caused by the glass on instrument dial,wavelet transformation with different parameters in the fan shape region when its average angle changes is used to abstract the informationof needle and graduation. The geometric structure characteristic space is built to calculate thecenter and the rotated angle of the tested image relative to the standard premodel. Then wecan recognize the readings through looking up the relative table between the needle angle andreadings. Satisfactory experimental results have been obtained by using the above system torecognize the readings of cockpit instrument dial image which are grabbed outdoors with noiseand reflection,the recognition precision is higher than that by manual operation and the systemis realized on main line measurement.